Veröffentlichungsnummer | US5291137 A |

Publikationstyp | Erteilung |

Anmeldenummer | US 07/970,332 |

Veröffentlichungsdatum | 1. März 1994 |

Eingetragen | 2. Nov. 1992 |

Prioritätsdatum | 2. Nov. 1992 |

Gebührenstatus | Bezahlt |

Auch veröffentlicht unter | US5381092 |

Veröffentlichungsnummer | 07970332, 970332, US 5291137 A, US 5291137A, US-A-5291137, US5291137 A, US5291137A |

Erfinder | Robert Freedman |

Ursprünglich Bevollmächtigter | Schlumberger Technology Corporation |

Zitat exportieren | BiBTeX, EndNote, RefMan |

Patentzitate (16), Nichtpatentzitate (22), Referenziert von (123), Klassifizierungen (7), Juristische Ereignisse (5) | |

Externe Links: USPTO, USPTO-Zuordnung, Espacenet | |

US 5291137 A

Zusammenfassung

A nuclear magnetic resonance (NMR) logging system includes a NMR logging tool adapted to be disposed in a borehole and a processing system adapted to be disposed on the surface of the borehole, the logging tool including a microcomputer, the processing system including a computer; and a signal processing system in accordance with the present invention having a first part stored in the downhole microcomputer of the logging tool and a second part stored in the surface oriented computer of the processing system. In the downhole microcomputer, using the first part of the signal processing system, a plurality of signal plus noise amplitudes are produced from a set of measured spin echo data, and a plurality N_{w} of the window sums I_{m},m+1 are produced from the signal plus noise ampltudes, the window sums being transmitted uphole to the processing system. By producing the window sums, the first part of the signal processing system compresses and eliminates redundant information from the measured spin echo data. In the surface oriented processing system, using the second part of the signal processing system, the plurality N_{w} of the window sums I_{m},m+1 are used to generate new data which is ultimately recorded on a new output record medium. Spectral amplitudes {a_{l} }, determined from the window sums, are used by the processing system to compute three log outputs and signal distributions presented as color maps. Three standard deviations are also computed. Log quality curves including signal phase and RMS noise are also computed.

Ansprüche(30)

1. A signal processing system adapted for extracting specific formation evaluation information from a set of measured spin echo data acquired from a logging tool disposed in a borehole, comprising:

first means adapted to be disposed within said logging tool in said borehole for compressing and eliminating redundant information from said set of measured spin echo data and generating a plurality of window sums, where the number of said plurality of window sums is less than the number of said set of measured spin echo data, said plurality of window sums adapted to be transmitted uphole from said logging tool;

second means adapted to be disposed on a surface of said borehole, connected to said logging tool and responsive to the plurality of window sums transmitted uphole from said logging tool for generating said specific formation evaluation information; and

means for recording said specific formation evaluation information on an output record medium.

2. The signal processing system of claim 1, wherein said first means generates a plurality of signal plus noise amplitudes A_{j}.sup.(+) from said set of measured spin echo data, where the number of said plurality of signal plus noise amplitudes is less than the number of said set of measured spin echo data,

said first means generating said plurality of window sums (I_{m},m+1) from said plurality of signal plus noise amplitudes A_{j}.sup.(+), where the number of said plurality of window sums is less than the number of said plurality of signal plus noise amplitudes,

whereby said first means compresses and eliminates redundant information from said set of measured spin echo data when said plurality of window sums (I_{m},m+1) is generated from said measured spin echo data.

3. The signal processing system of claim 2, wherein said specific formation evaluation information comprises:

a set of information including porosity, porosity standard deviation, free fluid standard deviation, and measurement diagnostics,

said set of information being recorded on said output record medium.

4. The signal processing system of claim 2, wherein said specific formation evaluation information comprises:

a set of information including porosity, spin spin relaxation time distributions, and continuous permeability logs,

said set of information being recorded on said output record medium.

5. A nuclear magnetic resonance logging system, comprising:

a logging tool adapted to be disposed in a wellbore, said logging tool including,

antenna means for sensing induced voltages representative of a plurality of magnetic moments associated with a plurality of protons precessing in a formation traversed by said wellbore and generating a plurality of spin echo receiver voltage pulses representative of the magnetic moments, amplitude generating means responsive to said plurality of spin echo receiver voltage pulses for generating a plurality (J) of inphase (R_{j}) amplitudes and a plurality (J) of quadrature (X_{j}) amplitudes, the plurality of inphase amplitudes and the plurality of quadrature amplitudes totalling 2J amplitudes, and

first means responsive to the inphase (R_{j}) and quadrature (X_{j}) amplitudes for generating a plurality (N_{w}) of window sums (I_{m},m+1), the plurality (N_{w}) of window sums being less in number than said 2J amplitudes;

processing means adapted to be disposed at a surface of said wellbore and responsive to said plurality of window sums generated by the first means of said logging tool for generating log outputs and signal distributions; and

output record generating means for recording said log outputs and said signal distributions representing color maps on an output record medium.

6. The logging system of claim 5, wherein said first means generates an estimate of RMS noise in response to the inphase and quadrature amplitudes,

said processing means determining a plurality of standard deviations in accordance with the estimate of RMS noise,

said output record generating means recording said plurality of standard deviations on said output record medium.

7. The logging system of claim 6, wherein said processing means determines a logarithm of a likelihood function from the estimate of RMS noise and further determines a set of spectral amplitudes {a_{1} } from the logarithm of the likelihood function,

said processing means generating said log outputs and said signal distributions in response to said spectral amplitude.

8. The logging system of claim 5, wherein said first means:

estimates a signal phase (theta) in response to the inphase and quadrature amplitudes generated by the amplitude generating means;

determines a plurality of signal plus noise amplitudes A_{j}.sup.(+) from said plurality of inphase amplitudes, said plurality of quadrature amplitudes, and said signal phase, said plurality of signal plus noise amplitudes being comprised of a plurality of groups of signal plus noise amplitudes; and

generates said plurality of window sums from the respective plurality of groups of signal plus noise amplitudes.

9. The logging system of claim 8, wherein said first means determines a plurality of amplitudes A_{j}.sup.(-) from said plurality of inphase amplitudes, said plurality of quadrature amplitudes, and said signal phase,

said first means generating an estimate of RMS noise from said plurality of amplitudes A_{j}.sup.(-).

10. The logging system of claim 9, wherein said processing means determines a plurality of standard deviations in accordance with the estimate of RMS noise,

said output record generating means recording said plurality of standard deviations on said output record medium.

11. The logging system of claim 10, wherein said processing means determines a logarithm of a likelihood function from the estimate of RMS noise and further determines a set of spectral amplitudes {a_{1} } from the logarithm of the likelihood function,

said processing means generating said log outputs and said signal distributions in response to said spectral amplitude.

12. A method of transmitting data from a well tool disposed in a wellbore to a processing system disposed at a surface of said wellbore, said well tool including an antenna for sensing information and generating a plurality of inphase amplitudes and a plurality of quadrature amplitudes in response to said information, comprising the steps of:

in said well tool, estimating a signal phase from the inphase and quadrature amplitudes;

determining a plurality of values A_{j}.sup.(+) from the signal phase, the plurality of inphase amplitudes, and the plurality of quadrature amplitudes;

sub-dividing said plurality of values into groups thereby producing a plurality of groups of said values, where the plurality of groups is less than the plurality of values;

generating a window sum value for each group of the plurality of groups of said values, thereby producing a plurality of window sums, where the plurality of window sums is equal to the plurality of groups of said values; and

transmitting the plurality of window sums uphole from the well tool disposed in said wellbore to the processing system disposed at the surface of said wellbore.

13. In a logging system including a well tool adapted to be disposed within a wellbore and a processing system disposed at a surface of said wellbore, a method of generating an output record medium illustrating parameters representative of the quality of a reservoir traversed by said wellbore, comprising the steps of:

(a) in said well tool, sensing induced voltages representative of a plurality of magnetic moments associated with a plurality of protons precessing in said reservoir traversed by said wellbore and generating a plurality of spin echo receiver voltage pulses representative of the magnetic moments;

(b) generating a plurality (J) of inphase (R_{j}) amplitudes and a plurality (J) of quadrature (X_{j}) amplitudes in response to said plurality of spin echo receiver voltage pulses, the plurality of inphase amplitudes and the plurality of quadrature amplitudes totalling 2J amplitudes, and

(c) generating a plurality (N_{w}) of window sums (I_{m},m+1) in response to the inphase (R_{j}) and quadrature (X_{j}) amplitudes, the plurality (N_{w}) of window sums being less in number than said 2J amplitudes;

(d) in said processing system, generating log outputs and signal distributions representing color maps in response to said plurality of window sums; and

(e) recording said log outputs and said signal distributions representing color maps on an output record medium.

14. The method of claim 13, wherein the generating step (c) comprises the step of:

(f) generating an estimate of RMS noise and said plurality (N_{w}) of window sums (I_{m},m+1) in response to said inphase (R_{j}) and quadrature (X_{j}) amplitudes.

15. The method of claim 14, wherein the generating step (d) comprises the step of:

(g) generating a plurality of standard deviations in response to the estimate of RMS noise, said log outputs and said signal distributions being generated in response to said estimate of RMS noise and said plurality of window sums.

16. The method of claim 15, wherein the recording step (e) comprises the step of:

(h) recording said plurality of standard deviations in addition to said log outputs and said signal distributions representing color maps on said output record medium.

17. The method of claim 14, wherein the generating step (d) comprises the step of:

(i) generating a plurality of standard deviations in response to the estimate of RMS noise;

(j) determining a logarithm of a likelihood function from the estimate of RMS noise;

(k) further determining a set of spectral amplitudes {a_{1} } from the logarithm of the likelihood function; and

(l) generating said log outputs and said signal distributions in response to said spectral amplitude.

18. The method of claim 17, wherein the recording step (e) comprises the step of:

(m) recording said plurality of standard deviations in addition to said log outputs and said signal distributions representing color maps on said output record medium.

19. The method of claim 13, wherein said generating step (c) comprises the steps of:

estimating a signal phase (theta) in response to the inphase and quadrature amplitudes;

determining a plurality of signal plus noise amplitudes A_{j}.sup.(+) from said plurality of inphase amplitudes, said plurality of quadrature amplitudes, and said signal phase, said plurality of signal plus noise amplitudes being comprised of a plurality of groups of signal plus noise amplitudes; and

generating said plurality of window sums from the respective plurality of groups of signal plus noise amplitudes.

20. The method of claim 19, wherein said generating step (c) further comprises the steps of:

determining a plurality of amplitudes A_{j}.sup.(-) from said plurality of inphase amplitudes, said plurality of quadrature amplitudes, and said signal phase, and

generating an estimate of RMS noise from said plurality of amplitudes A_{j}.sup.(-).

21. The method of claim 20, wherein the generating step (d) comprises the step of:

determining a plurality of standard deviations in accordance with the estimate of RMS noise.

22. The method of claim 21, wherein the generating step (d) further comprises the steps of:

determining a logarithm of a likelihood function from the estimate of RMS noise;

determining a set of spectral amplitudes {a_{1} } from the logarithm of the likelihood function; and

generating said log outputs and said signal distributions in response to said spectral amplitude.

23. The method of claim 22, wherein the recording step (e) comprises the step of:

recording said plurality of standard deviations in addition to said log outputs and said signal distributions representing color maps on said output record medium.

24. A system for receiving induced voltages associated with a material representative of a plurality of magnetic moments associated with a plurality of protons precessing in said material and generating an output record medium for recording information pertaining to a set of characteristics of said material, comprising:

means responsive to said induced voltages for generating a plurality of spin echo receiver voltage pulses representative of the magnetic moments;

amplitude generating means responsive to said plurality of spin echo receiver voltage pulses for generating a plurality (J) of inphase (R_{j}) amplitudes and a plurality (J) of quadrature (X_{j}) amplitudes, the plurality of inphase amplitudes and the plurality of quadrature amplitudes totalling 2J amplitudes;

first means responsive to the inphase (R_{j}) and quadrature (X_{j}) amplitudes for generating a plurality (N_{w}) of window sums (I_{m},m+1), the plurality (N_{w}) of window sums being less in number than said 2J amplitudes;

processing means responsive to said plurality of window sums generated by the first means for generating log outputs and signal distributions; and

output record generating means for recording said log outputs and said signal distributions on said output record medium.

25. The system of claim 24, wherein said first means generates an estimate of RMS noise in response to the inphase and quadrature amplitudes,

said processing means determining a plurality of standard deviations in accordance with the estimate of RMS noise,

said output record generating means recording said plurality of standard deviations on said output record medium.

26. Th system of claim 25, wherein said processing means determines a logarithm of a likelihood function from the estimate of RMS noise and further determines a set of spectral amplitudes {a_{1} } from the logarithm of the likelihood function,

said processing means generating said log outputs and said signal distributions in response to said spectral amplitude.

27. The system of claim 24, wherein said first means:

estimates a signal phase (theta) in response to the inphase and quadrature amplitudes generated by the amplitude generating means;

determines a plurality of signal plus noise amplitudes A_{j}.sup.(+) from said plurality of inphase amplitudes, said plurality of quadrature amplitudes, and said signal phase, said plurality of signal plus noise amplitudes being comprised of a plurality of groups of signal plus noise amplitudes; and

generates said plurality of window sums from the respective plurality of groups of signal plus noise amplitudes.

28. The system of claim 27, wherein said first means determines a plurality of amplitudes A_{j}.sup.(-) from said plurality of inphase amplitudes, said plurality of quadrature amplitudes, and said signal phase,

said first means generating an estimate of RMS noise from said plurality of amplitudes A_{j}.sup.(-).

29. The system of claim 28, wherein said processing means determines a plurality of standard deviations in accordance with the estimate of RMS noise,said output record generating means recording said plurality of standard deviations on said output record medium.

30. The system of claim 29, wherein said processing means determines a logarithm of a likelihood function from the estimate of RMS noise and further determines a set of spectral amplitudes {a_{1} } from the logarithm of the likelihood function,said processing means generating said log outputs and said signal distributions in response to said spectral amplitude.

Beschreibung

The subject matter of the present invention relates to a new method and apparatus for processing a signal which is output from a pulsed nuclear magnetism tool disposed in a wellbore thereby producing new output data representative of the formation traversed by the wellbore and for recording the new output data on an output record medium.

Repeated attempts have been made to use the principles of nuclear magnetic resonance by means of logging tools lowered through wellbores in oil exploration over the past several decades. It is recognized that particles of a formation having magnetic spin, for example atomic nuclei, protons, or electrons, have tendencies to align with a magnetic field which is imposed on the formation. Such a magnetic field may be naturally generated, as is the case with the earth's magnetic field (B_{E}) which has an intensity of approximately 0.5 gauss in areas of the globe where boreholes are typically drilled. Any given particle in a formation is additionally influenced by localized magnetic fields associated with nearby magnetic particles, other paramagnetic materials, and the layer of ions which typically line pore walls of certain types of formations such as shales. These localized fields tend to be inhomogeneous, while the earth's magnetic field is relatively homogeneous.

A nuclear magnetic resonance (NMR) logging tool apparatus, adapted to be disposed in a borehole, produces a static and a substantially homogeneous magnetic field focussed into a formation on one side of the logging tool. By directing and configuring the combined magnetic fields of a plurality of magnets, a region, remote from the plurality of magnets, is introduced wherein the special field gradient substantially vanishes, thereby insuring that the field is highly homogeneous throughout that region. In a preferred form, the magnets are mounted within a metallic skid or logging pad, the static magnetic field is directed through the face of the pad into an adjacent formation, and the region of substantially homogeneous field is situated in a volume of formation behind the mudcake layer which typically lines borehole walls. A homogeneous magnetic field, several hundred times stronger than the earth's magnetic field, can be thus imposed, or "focused", on a volume of formation in situ.

Reference may be had to U.S. Pat. No. 4,933,638 issued Jun. 12, 1990 to Kenyon et al (hereinafter termed, the "Kenyon et al patent" or "Kenyon et al") for details of such a nuclear magnetic resonance (NMR) logging tool apparatus, which patent is incorporated herein by reference. In the Kenyon et al patent, an RF antenna is mounted on the outside of the structure of the pad so that the pad serves as a natural shield against any signals which may be generated by resonant conditions behind the pad, particularly those potentially strong resonance signals from borehole fluid. In the preferred form, the antenna is configured to focus its signals radially outwardly from the pad face, into the volume of formation having the homogeneous field, thereby additionally reducing the distortion of measured signals from borehole effects.

All such nuclear magnetic resonance logging tool apparatus, when disposed in a borehole, are electrically connected to a computing apparatus disposed at the surface of the borehole. The computing apparatus stores a signal processing software therein, the software in conjunction with the hardware of the computing apparatus producing a plurality output data representative of the characteristics of the formation traversed by the borehole when the software is executed by the hardware while utilizing a set of input data which was developed by the logging tool disposed in the borehole.

While the prior art nuclear magnetic resonance logging tool of Kenyon et al, and its associated signal processing software, is capable of determining formation characteristics with sufficient accuracy and dependability, it has been found useful to improve the performance and accuracy of such logging tool, especially in view of the inherent difficulties of making nuclear magnetic resonance (NMR) measurements in boreholes.

One very important improvement in the performance of the Kenyon et al logging tool can be made to the signal processing software used by Kenyon et al. Several approaches to spectral decomposition or signal processing of NMR data have been reported. Spectral decomposition is a signal processing method that determines from NMR spin-echo signals in rocks, the individual amplitude components of the multi-exponential signal. These individual components correspond to different pore sizes in the rock.

A first approach is reported by Kenyon, W. E., Howard, J. J., Sezginer, A. Straley, C., Matteson, A., Horkowitz, R., and Erlich, R., in an article entitled "Poresize Distribution and NMR in Microporous Cherty Sandstones", Trans of the SPWLA of the 30th Ann. Logging Symp., Paper LL, Jun. 11-14, 1989, this first approach being further set forth in an article entitled "A NMR Technique for the Analysis of Pore Structure: Determination of Continuous Pore Size Distributions"; Gallegos, D. P. and Smith, D. M.; J. of Colloid and Interface Science, V. 122, No. 1, pp. 143-153, Mar. 1988, the disclosures of which are incorporated by reference into this specification.

A second approach is set forth in an article entitled "Problems in Identifying Multimodal Distributions of Relaxation Times for NMR in Porous Media", Magnetic Resonance Imaging, Vol 9, pp. 687-693 (1991), the disclosure of which is incorporated by reference into this specification.

The aforementioned approaches are computationally too intensive to be done in real time by a logging truck computer. They also do not compress the data which is needed to limit the telemetry requirements for sending data uphole. In addition, the NMR logging tool should first acquire downhole spin-echo measurements in earth formations penetrated by a borehole and then, secondly, generate a set of detailed formation evaluation information. However, this type of detailed formation evaluation information was previously obtainable only from costly laboratory analysis of conventional core data. Therefore, a signal processing method and apparatus is needed which is adapted to extract this formation evaluation information from the measured spin-echos. This signal processing method and apparatus must be capable of providing a spectral decomposition of the measured data, and it must be efficient and robust for real time processing of measured data during the acquisition of the data by an NMR logging tool moving in the borehole. Moreover, it is desirable to compress the data by elimination of redundant information. This reduces the telemetry requirements of the NMR tool, which is important if the tool is run in combination with other logging tools.

Accordingly, it is a primary object of the present invention to provide a new signal processing method and apparatus which is capable of extracting a new set of specific formation evaluation information from measured spin-echo data acquired from a nuclear magnetic resonance (NMR) logging tool which is disposed in a borehole and moves within such borehole.

It is a further object of the present invention to provide a new signal processing method and apparatus responsive to measured spin echo data for providing a spectral decomposition of the measured spin-echo data and subsequently generating a new set of specific formation evaluation information from the spectral decomposition, the signal processing method being efficient, flexible and robust for real time processing of the measured spin echo data during the acquisition of such data by a NMR logging tool moving axially within the borehole.

It is a further object of the present invention to provide a new signal processing method and apparatus responsive to a set of measured spin echo data for extracting specific formation evaluation information from the measured spin echo data acquired from a NMR logging disposed in a borehole, the signal processing method and apparatus including a first part stored downhole in a downhole microcomputer for compressing and eliminating redundant information from the set of measured spin echo data and generating a plurality of window sums, where the number of the plurality of window sums is much less than the number of the set of measured spin echo data, the plurality of window sums being transmitted uphole, and a second part stored uphole in a surface oriented computer system and responsive to the plurality of window sums received from the downhole microcomputer for generating a new set of specific formation evaluation information representative of characteristics of a formation traversed by the borehole.

It is a further object of the present invention to provide the new signal processing method and apparatus which generates the new set of specific formation evaluation information representative of characteristics of the formation traversed by the borehole, which new set of specific formation evaluation information includes total NMR porosity, free fluid porosity, bound fluid porosity, spin-spin (T2) relaxation time distributions which are related to pore size distributions in the formation, and continuous permeability logs.

It is a further object of the present invention to provide the new signal processing method and apparatus which generates the new set of specific formation evaluation information, which specific formation evaluation information includes the following information: total NMR porosity, free fluid porosity, porosity standard deviation, free fluid standard deviation, and measurement diagnostics including RMS noise estimate and signal phase.

It is a further object of the present invention to provide the new signal processing method and apparatus which provides the aforementioned new set of specific formation evaluation information and which generates a new output record medium illustrating the specific formation evaluation information, the new output record medium comprising a plurality of logs, each log illustrating a particular one of the aforementioned set of specific formation evaluation information and measurement diagnostics as a function of depth in the borehole.

In accordance with these and other objects of the present invention, a nuclear magnetic resonance (NMR) logging system includes: (1) a NMR logging tool adapted to be disposed in a borehole, the logging tool including a downhole microcomputer; (2) a processing system adapted to be disposed at the surface of the borehole and electrically connected to the logging tool for processing signals received from the logging tool, the processing system including a computer having a memory; and (3) a signal processing method and apparatus embodied in the form of a software package having a first part stored downhole in the downhole microcomputer of the logging tool for compressing and eliminating redundant information from received and measured spin echo data and generating compressed information; and a second part stored uphole in the surface oriented computer of the processing system and responsive to the compressed information for generating the new set of specific formation evaluation information and for further generating an output record medium which illustrates the specific formation evaluation information.

In the logging tool disposed in the borehole, a receiving antenna measures voltages induced by the precession of the magnetic moments of individual protons in the volumes of the formation traversed by the borehole and generates a plurality of spin-echo receiver voltage pulses representative of the magnetic moments. An electronics cartridge, which includes a microcomputer, begins processing the receiver voltage pulses by integrating each of the spin-echo receiver voltage pulses over a time interval, there being a total of J time intervals, each time interval being centered about a time t_{j} =j delta, where j=1, 2, . . . , J. The integrated signals are recorded as spin-echo inphase (R_{j}) and quadrature (X_{j}) amplitudes, time series channels or waveforms. In the DESCRIPTION OF THE PREFERRED EMBODIMENT, the symbols "R_{j} " and "X_{j} " will be written without a "tilde symbol overbar", whereas the same symbols in the DETAILED DESCRIPTION will be written with the tilde symbol overbar.

In the microcomputer, using the first part of the signal processing method and apparatus in accordance with the present invention, a signal phase (theta) is estimated from the 2J spin-echo in-phase (R_{j}) and quadrature (X_{j}) amplitudes associated with the 2J spin-echo receiver voltage pulses. Then, the in-phase (R_{j}) amplitude, quadrature (X_{j}) amplitude, and the signal phase (theta) associated with each of the spin-echo receiver voltage pulses are combined to produce a signal plus noise amplitude A_{j}.sup.(+). A plurality of the signal plus noise amplitudes A_{j}.sup.(+) are disposed within a time window, there being a plurality of time windows. A sum of the plurality of signal plus noise amplitudes A_{j}.sup.(+) within each time window produces a window sum I_{m},m+1 ; and, since there are a plurality of time windows, there are a plurality of window sums I_{m},m+1. Consequently, the downhole microcomputer of the NMR logging tool disposed in the borehole generates a plurality of the window sums I_{m},m+1, one for each time window, and transmits the plurality of the window sums uphole to the processing system located at the surface of the borehole. Note that each window sum I_{m},m+1 itself represents a reduced set of data, since each window sum I_{m},m+1 is the sum of a plurality of signal plus noise amplitudes A_{j}.sup.(+). As a result, the telemetry requirements needed by the logging tool to transmit the plurality of window sums I_{m},m+1 uphole to the processing system located at the wellbore surface are substantially reduced. In addition to producing the signal plus noise amplitude A_{j}.sup.(+), the downhole microcomputer also produces a set of "J" amplitudes A_{j}.sup.(-). These amplitudes A_{j}.sup.(-) are used by the downhole microcomputer for estimating an RMS noise, which is defined to be the square root of "psi", where "psi" is the noise power. The RMS noise is also transmitted uphole to the processing system disposed at the wellbore surface.

In the processing system disposed at the wellbore surface, using the second part of the signal processing method and apparatus in accordance with the present invention, the RMS noise is used to compute three standard deviations: the standard deviation "sigma(phi_{nmr})", the standard deviation "sigma(phi_{f})", and the standard deviation "sigma(T_{2},log)". These standard deviations are used to generate the new output record medium in accordance with the present invention. The RMS noise is also used to compute the dimensionless parameter "gamma", the use of which is discussed below. As noted earlier, the plurality N_{w} of the window sums I_{m},m+1 are transmitted uphole by the downhole computer. The processing system located at the wellbore surface includes a surface computer. The surface computer receives the plurality N_{w} of the window sums I_{m},m+1 ; and, using the plurality N_{w} of window sums I_{m},m+1 and the aforementioned dimensionless parameter "gamma", the surface computer determines a logarithm of the likelihood function (-1n L) for the N_{w} window sums set forth in equation 42 in the detailed description of the preferred embodiment. The logarithm of the likelihood function (-1n L) of equation 42 is a function of a set of spectral amplitudes "a_{1} ", where such spectral amplitudes are determined by minimization of equation 42. The spectral amplitudes "a_{1} " are used by the surface computer: (1) to compute three log outputs, phi_{nmr}, phi_{f}, and T_{2},log, and (2) to compute signal distributions P_{a} (logT_{2}) which are represented by color maps.

The new output record medium in accordance with the present invention is generated using the three log outputs, the signal distributions for color maps, and the three standard deviations.

Further scope of applicability of the present invention will become apparent from the detailed description presented hereinafter. It should be understood, however, that the detailed description and the specific examples, while representing a preferred embodiment of the present invention, are given by way of illustration only, since various changes and modifications within the spirit and scope of the invention will become obvious to one skilled in the art from a reading of the following detailed description.

A full understanding of the present invention will be obtained from the detailed description of the preferred embodiment presented hereinbelow, and the accompanying drawings, which are given by way of illustration only and are not intended to be limitative of the present invention, and wherein:

FIG. 1 illustrates a nuclear magnetic resonance (NMR) logging system including a NMR logging tool disposed in a wellbore and a processing system disposed at the wellbore surface for processing signals transmitted uphole by the logging tool;

FIG. 2 illustrates a sketch of the logging tool in the wellbore producing a static magnetic field B0 into a formation traversed by the wellbore;

FIGS. 3-5 illustrate sections of the formation adjacent the logging tool in the wellbore and the effects of the magnetic field B0 on the composite magnetic moment vector M representative of the sum of a plurality of individual magnetic moments associated with a corresponding plurality of protons disposed within elements of the formation;

FIGS. 6-7 illustrate the effects on the proton's composite magnetic moment M of the presence of two magnetic fields: the static magnetic field B0 and a radio frequency magnetic field B_{x90}, the direction of which is transverse to the static magnetic field B0 and along the x-axis, the field B_{x90} being applied along the x-axis for a duration "T90" which causes a 90 degree rotation of the composite magnetic moment vector M about the x-axis;

FIGS. 7a-7b illustrate the precession of the individual magnetic moments vector m1 and the effects on this magnetic moments vector m1 of the presence of two other magnetic fields: the static magnetic field B0 and a radio frequency magnetic field By_{180}, the direction of which is transverse to the magnetic field B0 and is directed along the y-axis;

FIGS. 8a-8m illustrate the effects on the precession rate of an individual magnetic moment "A_{n} " which is located within a stronger field strength of the static magnetic field B0, relative to another individual magnetic moment "A_{f} " which is located within a weaker field strength of the static magnetic field B0, since the field B0 is non-homogeneous and the individual magnetic moments precess at different rates about the z-axis, the direction of the magnetic field B0 vector, and the effects on the magnetic moments A_{n} and A_{f} of a further magnetic field pulse B_{y180} which "rotates" the magnetic moments A_{n} and A_{f} 180 degrees about the y-axis thereby causing the magnetic moments A_{n} and A_{f} to refocus and to produce the spin-echo receiver voltage pulses of FIG. 9b;

FIGS. 9a-9b illustrate the spin-echo transmitter and receiver voltage pulses produced by the logging tool in time relation to: (1) a first magnetic field pulse which is directed along the x-axis of FIG. 6 and has a pulse duration of T90 (the first magnetic field pulse being hereinafter termed "B_{x90} " or "the 90 pulse"), and (2) a second magnetic field pulse directed along the y-axis and having a pulse duration of T180 which rotates all of the proton spins 180 degrees about the y-axis (the second magnetic field pulse being hereinafter termed B_{y180});

FIG. 10 illustrates NMR logging tool disposed in a borehole including an electronics cartridge which stores the first part of the signal processing method and apparatus in accordance with the present invention, and the processing system disposed on the surface of the borehole which stores the second part of the signal processing method and apparatus in accordance with the present invention;

FIG. 11 illustrates a more detailed construction of the surface oriented processing system of FIG. 10 which stores the second part of the signal processing method and apparatus of the present invention;

FIGS. 12a, 12b1 and 12b2 illustrate a more detailed construction of the electronics cartridge of FIG. 10 which stores the first part of the signal processing method and apparatus of the present invention;

FIG. 13 illustrates a block diagram or flow chart of the first part of the signal processing method and apparatus of the present invention stored in the electronics cartridge of the NMR tool and the second part of the signal processing method and apparatus of the present invention stored in the surface oriented processing system of FIG. 10, which first part and second part of the signal processing method and apparatus is embodied in the form of software stored,, in a memory of the electronics cartridge and the surface oriented processing system, respectively;

FIG. 14 illustrates a plurality of the spin-echo signal plus noise amplitudes A_{j}.sup.(+) corresponding, respectively, to the plurality of spin-echo signals of FIG. 9b, each spin-echo signal plus noise amplitude A_{j}.sup.(+) being identified by an echo number, the echo numbers being divided into a plurality of time windows;

FIG. 15 illustrates a plurality of window sums I_{m},m+1 corresponding, respectively, to the plurality of time windows of FIG. 14;

FIG. 16 illustrates a new output record medium, comprising a plurality of new logs, generated by the processing system disposed at the, wellbore surface in accordance with another aspect of the present invention;

FIG. 16a illustrates a more detailed block diagram of the surface computer and the downhole computer of FIG. 12a utilized in conjunction with FIG. 13 for providing a functional description of the present invention;

FIG. 17 illustrates the standard deviation in porosity estimates versus the total number of echos for two values of "gamma";

FIG. 18 illustrates the standard deviation in porosity estimates versus the total number of echos for two values of t_{cp} ;

FIG. 19 illustrates the standard deviation in porosity estimates versus the total number of echos for two values of W:

FIG. 20 illustrates the standard deviation in porosity estimates versus the total number of echos for three values of T_{min} ;

FIG. 21 illustrates a signal distribution computed from station data at 1100 ft for two values of gamma differing by an order of magnitude;

FIG. 22 illustrates a signal distribution at 1125 ft exhibiting a two peak structure reflecting two disparate pore size distributions;

FIG. 23 illustrates a signal distribution at 1148 ft indicating a relatively poor reservoir quality rock;

FIG. 24 illustrates a signal distribution at 1200 ft indicating a relatively good quality permeable reservoir rock;

FIG. 25 illustrates a signal distribution at 1229 ft indicating essentially no bound fluid, and a long relaxation time; and

FIG. 26 illustrates a correction "delta phi" used in equation C.1 associated with a calculation of "delta phi".

This specification is divided into two parts:

(1) a Description of the Preferred Embodiment, which provides a general, more understandable description of the new signal processing method and apparatus of the present invention; and

(2) a Detailed Description of the Preferred Embodiment, which provides a more detailed description of the new signal processing method and apparatus of the present invention.

Referring to FIG. 1, a nuclear magnetic resonance (NMR) logging system is illustrated, the NMR logging system including a NMR logging tool 13 disposed in a wellbore and a processing system 7a disposed at the wellbore surface for processing signals transmitted uphole by the logging tool.

In FIG. 1, a borehole 10 is shown adjacent to formations 11, 12, the characteristics of which are to be determined. Within borehole 10 there is shown a logging tool 13 connected via a wireline 8 to surface equipment 7. The surface equipment 7 includes a processing system 7a which stores therein a signal processing method and apparatus embodied in the form of software. The processing system 7a will be discussed in more detail with reference to FIGS. 12 and 13 of the drawings. Tool 13 preferably has a face 14 shaped to intimately contact the borehole wall, with minimal gaps or standoff. The tool 13 also has a retractable arm 15 which can be activated to press the body of the tool 13 against the borehole wall during a logging run, with the face 14 pressed against the wall's surface. Although tool 13 is shown in the preferred embodiment of FIG. 1 as a single body, the tool may obviously comprise separate components such as a cartridge, sonde or skid, and the tool may be combinable with other logging tools as would be obvious to those skilled in the art. Similarly, although wireline 8 is the preferred form of physical support and communicating link for the invention, alternatives are clearly possible, and the invention can be incorporated in a drill stem, for example, using forms of telemetry which may not require a wireline. The formations 11, 12 have distinct characteristics such as formation type, porosity, permeability and oil content, which can be determined from measurements taken by the tool. Deposited upon the borehole wall of formations 11, 12 is typically a layer of mudcake 16 which is deposited thereon by the natural infiltration of borehole fluid filtrate into the formations. In the preferred embodiment shown in FIG. 1, tool 13 comprises a first set magnet array 17 and an antenna 18 positioned between the array 17 and the wall engaging face 14. Magnet array 17 produces a static magnetic field B0 in all regions surrounding the tool 13. The antenna 18 produces, at selected times, an oscillating radio frequency magnetic field Bl (previously denoted as "B_{x90} " and B_{y180} ") which is focussed into formation 12, and is superposed on the static field B0 within those parts of formation opposite the face 14. The field Bl is perpendicular to the field B0. The Volume of Investigation, 19, of the tool for the first set magnet array 17 shown in dotted lines FIG. 1, is a vertically elongated region directly in front of tool face 14 in which the magnetic field produced by the magnet array 17 is substantially homogeneous and the spatial gradient thereof is approximately zero. The tool 13 may also comprise, as an option, a second set magnet array 20 and an antenna 21 positioned between the array 20 and the wall engaging face 14. Magnet array 20 produces another static magnetic field B0 in all regions surrounding the tool 13. The antenna 21 produces, at selected times, an oscillating radio frequency magnetic field Bl which is again focussed into formation 12, and is superposed on the static field B0 within those parts of formation opposite the face 14. The Volume of Investigation 22 of the tool for the second set magnet array 20, shown in dotted lines in FIG. 1, is a vertically elongated region directly in front of tool face 14 in which the magnetic field produced by the magnet array 20 is substantially homogeneous and the spatial gradient thereof is approximately zero. Due to the particular magnet arrangement for the second set magnet array 20, the Volume of Investigation 22 is at a depth in the formation 12 which is greater than the depth at which the Volume of Investigation 19 is located. A prepolarizing magnet 23, shown in dotted lines, may be positioned directly above the array 17 in a modified embodiment of the invention in accordance with the teachings of the aforementioned Kenyon, et al patent. An electronics cartridge 24 is positioned above the magnet 23. The electronics cartridge 24 includes a downhole microcomputer. The electronics cartridge 24, including the downhole microcomputer, will be discussed in more detail with reference to FIGS. 12a-12b of the drawings.

In operation, referring to FIG. 1, the tool 13 makes a measurement in the Volume of Investigation 19 by magnetically reorienting the nuclear spins of particles in formation 12 with a pulse of oscillating magnetic field Bl (previously denoted as "B_{x90} " and B_{y180} "), and then detecting the precession of the tipped particles in the static, homogeneous field BO within the Volume of Investigation 19, over a period of time. As seen in FIG. 1, this Volume of Investigation does not overlap the surface of the wall engaging face 14 as in some previous logging tools, and does not overlap the mudcake 16 on the borehole wall. In a pulse echo type of measurement, a pulse of RF current is passed through the antenna 18 to generate a pulse of RF field Bl where the RF frequency is selected to resonate only hydrogen nuclei subjected to a static magnetic field strength equal or nearly equal to the field B0 within the Volume of Investigation 19. The signals induced in antenna 18 subsequent to the RF pulse Bl represent a measurement of nuclear magnetic precession and decay within the Volume, automatically excluding any undesirable contributions from the borehole fluid, mudcake, or surrounding formations where the field strength of B0 is substantially different. The tool 13 makes a measurement in the Volume of Investigation 22 in the same manner discussed above with respect to the Volume of Investigation 19 but utilizing the second set magnet array 20 and the antenna 21.

The general principles associated with nuclear magnetic resonance well logging, utilized by the NMR logging system of FIG. 1, will be discussed in the following paragraphs with reference to FIGS. 2 through 10 of the drawings.

In FIG. 2, a NMR tool 13 contacts a wall of the borehole 10. A magnet disposed within the tool 13 generates a static magnetic field B0, which field B0 is directed toward a volume of investigation 19 disposed within a portion of the formation 12 traversed by the borehole 10.

FIG. 3 illustrates a cross section 19a of the volume of investigation 19 of FIG. 2. The cross section 19a includes a plurality of pores 19al, each of which contain oil and/or water, the oil and/or water being further comprised of a plurality of protons, each proton having a magnetic spin or magnetic moment (u_{i}) identified in FIG. 3 by the arrow A shown in FIG. 3. Since there are a plurality of protons in each of the pores 19al, there are a corresponding plurality of magnetic moments (u_{1}, u_{2}, u_{3}, . . . , u_{n}) in each pore 19al, where "n" equals the number of protons in each pore.

FIG. 4 illustrates a plurality of magnetic moments "u_{i} " (identified in FIG. 4 by a plurality of arrows "A" ) associated with a corresponding plurality of protons disposed within a particular one of the pores 19al of the volume of investigation 19 when the static magnetic field B0 is zero. Note that the magnetic moments (u_{i}) A all point in different directions. Therefore, a composite magnetic moment "M", defined to be the vector sum of all individual magnetic moments A (that is, M=u_{1} +u_{2} +u_{3} +. . . +u_{n}) is approximately zero.

However, FIG. 5 illustrates the same plurality of magnetic moments A of FIG. 4, but now the magnetic field B0 is not equal to zero. Note that, in FIG. 5, the magnetic moments (u_{i}) A align together in the same direction when the magnetic field B0 is not equal to zero. Therefore, the composite magnetic moment "M" is not approximately equal to zero, but rather, is equal to a sum which is defined to be the sum of all individual magnetic moments A. Since each individual magnetic moment A can be quantified by the term "u_{i} ", the composite magnetic moment "M" is equal to the sum of all individual magnetic moments "u_{i} ", where i=1,2,3 . . . ,n, associated with the "n" individual protons disposed within a pore space 19al associated with the volume of investigation 19, that is, M=u_{1} +u_{2} +u_{3} +. . . +u_{n}.

Referring to FIG. 6, assume that the composite magnetic moment "M" for the volume of investigation 19 of FIG. 3-5 is aligned along the "z" axis; assume further that the static magnetic field B0 is also aligned along the z-axis. Then, assume that an oscillating radio frequency magnetic field pulse "B1" or "B_{x90} " is applied along the x-axis, transverse to the z-axis.

Referring to FIG. 7, recalling the assumptions mentioned above with reference to FIG. 6, since the pulse duration of the oscillating magnetic field pulse "B_{x90} " is 90 degrees and is applied along the x-axis in the presence of the static magnetic field B0, the composite magnetic moment "M" rotates 90 degrees from the z-axis to the y-axis, as shown in FIG. 7.

Referring to FIGS. 7a-7b, the magnetic moment vector M is initially disposed along the y-axis as shown in FIG. 7, and begins to "precess" (or rotate) clockwise within the x-y plane, as shown in FIG. 7a. However, in FIG. 7b, if an oscillating magnetic field pulse B_{y180} is applied along the y-axis (for a period of time equivalent to a 180 degree pulse duration), the magnetic moment vector M rotates or flips 180 degrees from one quadrant within the x-y plane to another quadrant within the x-y plane, as illustrated by element numeral 30 in FIG. 7b. The significance of this concept will become clear with reference to FIGS. 8a-8l of the drawings.

Referring to FIG. 8a-8m, recall that the composite magnetic moment "M" is defined to be the vector sum of all the individual magnetic moments "u_{i} " (identified by the letter A) of all protons within a pore space 19al in FIG. 3, that is, M=SUMMATION u_{i}, where i=1, 2, . . . , n, or M=u_{1} +u_{2} +u_{3} +. . . +u_{n}. In FIGS. 7a-7b, it was shown that, in the presence of the static magnetic field B0, the composite magnetic moment vector M precessed in the x-y plane, then flipped to another quadrant in response to the B_{y180} oscillating magnetic field pulse.

Since the composite magnetic moment vector M is the sum of all individual magnetic moments "u_{i} ", in actuality, all of the individual magnetic moments "u_{i} ", being disposed within the static magnetic field B0, individually precess in the x-y plane; and, then, each individual magnetic moment "u_{i} " flips to another quadrant in response to the B_{y180} oscillating magnetic field pulse.

However, the static magnetic field B0 is not homogeneous, that is, some parts of the static magnetic field B0 are stronger in terms of field strength than other parts of the field B0. Therefore, if one individual magnetic moment "u_{1} " is disposed within a stronger part of the static magnetic field B0, and another individual magnetic moment "u_{2} " is disposed within a weaker part of the static magnetic field B0, the rate of precession or rotation within the x-y plane of the one magnetic moment "u_{1} " will be greater than the rate of precession or rotation within the x-y plane of the other magnetic moment "u_{2} ".

FIGS. 8a-8m illustrate the effects on the precession rate of the one individual magnetic moment "u_{1} " which is located within a stronger part of the field B0 relative to another individual magnetic moment "u_{2} " which is located within a weaker part of the field B0; furthermore, FIGS. 8a-8l illustrate the effects on the one individual magnetic moment "u_{1} " and the other individual magnetic moment "u_{2} " of a further, oscillating magnetic field pulse B_{y180} which "rotates" or "flips" the magnetic moments "u_{1} " and "u_{2} " 180 degrees about the y-axis thereby causing the magnetic moments "u_{1} " and "u_{2} " to refocus and to produce one of the spin-echo receiver voltage pulses of FIG. 9b.

In FIG. 8a, the plurality of individual magnetic moments "u_{i} " A (and thus, the composite magnetic moment M) rotate 90 degrees in the z-y plane, as shown in FIGS. 6-7 of the drawings, in response to the static, radio frequency magnetic field pulse "B_{x90} " which is applied along the x-axis for a time equivalent to a 90 degree pulse duration.

In FIG. 8b, the one individual magnetic moment "u_{1} ", the magnetic moment associated within one particular proton which is disposed within a stronger part of the magnetic field B0, precesses or rotates, in a clockwise direction, at a faster rate than does the other individual magnetic moment "u_{2} ", the other individual magnetic moment associated with another proton that is located within a weaker part of the magnetic field B0.

In FIGS. 8c and 8m, a further, oscillating magnetic field pulse B_{y180} is applied at a time t_{cp} after application of the magnetic field pulse B_{x90} (which is assumed to be applied at a time t=0) along the y-axis for a period of time equivalent to a pulse duration of 180 degrees; in response to the further magnetic field pulse B_{y180}, the individual magnetic moments "u_{1} " and "u_{2} " rotate or flip about the y-axis by an amount equal to 180 degrees thereby removing the two individual magnetic moments "u_{1} " and "u_{2} " from quadrant number 2 and locating the two individual magnetic moments "u_{1} " and "u_{2} " in quadrant number 1 of the x-y plane, as shown in FIG. 8c.

In FIG. 8d, both of the individual magnetic moments "u_{1} " and "u_{2} " continue to rotate in the clockwise direction but the one magnetic moment "u_{1} " continues to rotate or precess at a faster rate than does the other magnetic moment "u_{2} ".

In FIG. 8e, since, as shown in FIG. 8c, the two individual magnetic moments "u_{1} " and "u_{2} " flipped from quadrant 2 to quadrant 1 of the x-y plane in response to the oscillating magnetic field B_{y180}, the two individual magnetic moments "u_{1} " and "u_{2} " refocus (align together as one magnetic moment vector) thereby producing a first spin-echo signal (echo 1) at a time 2t_{cp}, which spin-echo signal is picked up by the antennas 18 or 21 of the NMR tool 13 of FIG. 1 and is transmitted uphole to the processing system 7a. The first spin-echo signal (echo 1) is shown in FIG. 9b of the drawings.

In FIG. 8f, the one magnetic moment "u_{1} " proceeds to precess ahead or in front of the other magnetic moment "u_{2} " since the rate of precession of the one magnetic moment "u_{1} " is still greater than the rate of precession of the other individual magnetic moment "u_{2} ".

In FIGS. 8g and 8m, a further oscillating magnetic field pulse B_{y180} is applied at a time 3t_{cp} along the y-axis for a period of time equivalent to a pulse duration of 180 degrees; in response to the further magnetic field pulse B_{y180}, the two individual magnetic magnetic moments "u_{1} " and "u_{2} " rotate or flip about the y-axis by an amount equal to 180 degrees (similar to the action depicted in FIG. 8c) thereby removing the two magnetic moments "u_{1} " and "u_{2} " from quadrant number 2 and locating the two magnetic moments "u_{1} " and "u_{2} " into quadrant number 1 of the x-y plane, as shown in FIG. 8g.

In FIG. 8h, the two individual magnetic moments "u_{1} " and "u_{2} " refocus again, similar to the action depicted in FIG. 8e, thereby producing a second spin-echo signal (echo 2) at a time 4t_{cp}, which second spin-echo signal is picked up by antennas 18 and 21 of the NMR tool 13 of FIG. 1 and transmitted uphole to the processing system 7a. FIG. 9b illustrates the echo 2, second spin-echo signal. In FIG. 8i, the one individual magnetic moment "u_{1} " proceeds ahead of the other individual magnetic moment "u_{2} " since the rate of precession of the one magnetic moment "u_{1} " is greater than the rate of precession of the other magnetic moment "u_{2} ". In FIGS. 8j and 8m, the two individual magnetic moments "u_{1} " and "u_{2} " flip or rotate 180 degrees about the y-axis from quadrant 2 to quadrant 1 of the x-y plane in response to application of the oscillating magnetic field pulse B_{y180}, applied along the y-axis at a time 5t_{cp}.

In FIG. 8k, the two individual magnetic moments "u_{1} " and "u_{2} " refocus again thereby producing a third spin-echo signal (echo 3) at a time 6t_{cp}, which third spin-echo signal (echo 3) is picked up by antennas 18 and 21 of NMR tool 13 and transmitted uphole to processing system 7a. The echo 3, third spin-echo signal is illustrated in FIG. 9b.

In FIG. 8l, the one individual magnetic moment "u_{1} " proceeds ahead of the other individual magnetic moment "u_{2} " since the rate of precession of the one magnetic moment "u_{1} " is greater than the rate of precession of the other magnetic moment "u_{2} ".

Therefore, three spin-echo signals have been produced, the first spin-echo signal (echo 1) being produced in connection with FIG. 8e, the second spin-echo signal (echo 2) being produced in connection with FIG. 8h, and the third spin-echo signal (echo 3) being produced in connection with FIG. 8k.

Referring to FIGS. 9a-9b, the aforementioned first and second spin-echo signals, echo 1 and echo 2, are illustrated in time relation to the oscillating magnetic field pu_{1} se B_{x90} (90 pulse) which is directed along the x-axis of FIG. 6 and has a pulse duration of 90 degrees, and to the further oscillating magnetic field pulse B_{y180} (180 pulse) directed along the y-axis and having a pulse duration of 180 degrees. FIG. 9b illustrates a plurality of spin-echo signals associated with either the (R_{j}) inphase channel or the (X_{j}) quadrature channel. When FIGS. 9a-9b are examined jointly with FIGS. 8a-8m, it is evident that the nth spin-echo signal is produced at a time 2(n)t_{cp} following application of the magnetic field pulses B_{y180} at (2n-1)t_{cp}, where n=1, 2, 3, . . . , J. The first spin-echo signal (echo 1) is generated after a time t_{cp} elapses following application of the first magnetic field pulse B_{y180} and the second spin-echo signal (echo 2) is generated after a time t_{cp} elapses following application of the second magnetic field pulse B_{y180}.

Referring to FIG. 10, a nuclear magnetic resonance (NMR) logging system is illustrated, the logging system including a NMR logging tool 13 disposed in a wellbore and surface equipment 7 in the form of a well logging truck 7 situated on the surface of the wellbore, the well logging truck 7 including a processing system 7a in the form of a computer 7a situated within the well logging truck 7.

In FIG. 10, the NMR logging tool 13 includes antennas 18 and 21 and an electronics cartridge 24 responsive to signals received by the antennas 18 and 21 for processing the signals before transmission of the signals uphole to the computer 7a in the well logging truck. The electronics cartridge 24 of the logging tool 13 stores a first part of the signal processing (software) method and apparatus in accordance with one aspect of the present invention, and the computer 7a stores a second part of the signal processing (software) method and apparatus in accordance with another aspect of the present invention.

In FIG. 11, the computer 7a includes a system bus 7a1, a processor 7a3 connected to the system bus 7a1 for generating new output data in accordance with one aspect of the present invention, a memory 7a4 connected to the system bus 7a1 for storing the second part of the signal processing (software) method and apparatus of the present invention, and a recorder 7a2 connected to the system bus for receiving the new output data from the processor and generating a new output record medium 7a2A, to be discussed with reference to FIG. 16 of the drawings, in accordance with another aspect of the present invention. The computer 7a may include or consist of any one of the following computer systems manufactured by Digital Equipment Corporation (DEC), Maynard, Mass.: (1) DEC VAX 6430, (2) DEC PDP-11, or (3) DEC Vaxstation 3100, or it may include any other suitable computer system.

Referring to FIGS. 12a, 12b1, and 12b2, a contruction of the electronics cartridge 24 of FIG. 10 is illustrated. A more detailed construction of the hardware associated with the NMR logging system of FIGS. 1, 12a, 12b1, and 12b2, and in particular, of the construction of the electronics cartridge 24, is set forth in a prior pending application entitled "Borehole Measurement of NMR Characteristics of Earth Formations", corresponding to allowed application Ser. No. 07/970,324 filed Nov. 2, 1992, the disclosure of which is incorporated by reference into this specification.

In FIG. 12a, the surface computer 7a is electrically connected via logging cable to the electronics cartridge 24 of FIG. 10. The electronics cartridge 24 includes a downhole computer 46. The downhole computer 46 is ultimately electrically connected to the antennas 18 and 21.

In FIGS. 12b1, the downhole computer 46 of electronics cartridge 24 includes a system bus 46b to which a processor 46c is electrically connected and a memory 46a is electrically connected.

In FIG. 12b2, the memory 46a stores the first part of the signal processing method and apparatus of the present invention.

Referring to FIG. 13, a flow chart of the first part and the second part of the signal processing (software) method and apparatus of the present is illustrated.

This specification is divided into a Description of the Preferred Embodiment, and a Detailed Description of the Preferred Embodiment. The flow chart of FIG. 13 and the following discussion is part of the Description of the Preferred Embodiment and provides a general discussion of the signal processing method and apparatus of the present invention. The Detailed Description of the Preferred Embodiment set forth below provides a more detailed discussion of the aforementioned signal processing method and apparatus of the present invention. In the following general discussion, occasional reference to equations and other specific description set forth in the Detailed Description of the Preferred Embodiment will be required.

The signal processing method and apparatus of the present invention, embodied in the form of a software package, includes two parts: a first part 46a1 stored in the memory 46a of the downhole computer 46 of electronics cartridge 24 of FIGS. 12a-12b2 and a second part 7a4A stored in the memory 7a4 of the well logging truck computer 7a of FIG. 10 situated at the uurface of the wellbore.

In FIG. 1, the receiving antennas 18 and 21 of logging tool 13 measure the magnetic moments of individual protons in the volumes 19 and 22 of the formation traversed by the borehole 10 and generate a plurality of spin-echo receiver voltage pulses (FIG. 9b) representative of the magnetic moments. The electronics cartridge 24 begins processing the two-channel receiver voltage pulses by integrating each of the spin-echo receiver voltage pulses over a time interval, there being a total of J time intervals, each time interval being centered about a time t_{j} =j delta, where j=1, 2, . . . ,J. The integrated signals are recorded as spin-echo inphase (R_{j}) and quadrature (X_{j}) amplitudes, time series channels or waveforms.

Referring to FIG. 13, the first part 46a1 of the signal processing method and apparatus of the present invention is illustrated.

In FIG. 13, the first part 46a1 of the signal processing method and apparatus of the present invention receives the aforementioned inphase (R_{j}) and quadrature (X_{j}) amplitudes, and a signal phase (theta) is estimated associated with these amplitudes, block a1A. Equation 9 of the Detailed Description set forth below provides the equation of the signal phase (theta) as a function of the inphase (R_{j}) and quadrature (X_{j}) amplitudes, as follows: ##EQU1## Since the inphase (R_{j}) amplitude, the quadrature (X_{j}) amplitude, and the signal phase (theta) is known for each spin-echo receiver voltage pulse, the signal plus noise amplitude A_{j} (+) and the amplitude A_{j}.sup.(-), for each spin-echo receiver voltage pulse, may now be determined, block a1B, by utilizing equations 22 and 23 of the Detailed Description, as follows:

A_{j}.sup.(+) =R_{j}cos θ+X_{j}sin θ, (22)

A_{j}.sup.(-) =R_{j}sin θ-X_{j}cos θ. (23)

The RMS noise, the square root of psi or "SQRT psi", can be estimated from the signal plus noise amplitude A_{j}.sup.(-), block alC, by utilizing a practical implementation of equation 31 of the Detailed Description, as follows:

1J SUMMATION (j=1 . . . J) (A_{j}.sup.(-))^{2}approximately =psi,

where equation (31) is set forth as follows:

<(A_{j}.sup.(-))^{2}>≃ψ, (31)

The window sum I_{m},m+1 can be computed from the signal plus noise amplitude A_{j}.sup.(+), block a1D, by utilizing equations 22 and 35 of the Detailed Description, as follows: ##EQU2##

As a result, when the downhole computer 46 of FIG. 12 executes the first part 46a1 of the signal processing software of FIG. 13, two outputs are generated: the window sum I_{m},m+1 which is determined from equation 35 and the RMS noise (SQRT psi) which is determined from equation 31.

The following paragraphs present a more detailed explanation of the window sum I_{m},m+1 determined from equation 35.

Referring to FIG. 14, an example of signal plus noise amplitudes A_{j}.sup.(+) that have been determined from spin-echo signals (echo 1, echo 2,...) like those shown schematically as either R or X channel spin-echo pulses in FIG. 9b is illustrated in FIG. 14. The spin-echo signal plus noise amplitudes A_{j}.sup.(+) are separated in time by "2t_{cp} " from each other.

Referring to FIGS. 14 and 15, the technique for determining a particular window sum (I_{m},m+1) from a plurality of signal plus noise amplitudes (A_{1}.sup.(+), A_{2}.sup.(+), A_{3}.sup.(+), A_{4}.sup.(+), . . . , A_{n}.sup.(+)) is illustrated.

Recalling that a window sum I_{m},m+1 is determined from the equation "I_{m},m+1 =SUMMATION A_{j}.sup.(+) ", as set forth in equation 35 of the Detailed Description of the Preferred Embodiment set forth below, referring to FIGS. 14 and 15, a first window sum "I_{1},2 ", where m=1, may be determined by summing a plurality of individual signal plus noise amplitudes A_{1}.sup.(+), A_{2}.sup.(+), A_{3}.sup.(+), A_{4}.sup.(+), . . . , A_{n}.sup.(+) which are disposed within a first time window 50 shown in FIG. 14. A second window sum I_{2},3 is determined in association with a second time window 52 and a third window sum I_{3},4 is determined in association with a third time window 54 in the same manner as indicated above by summing the associated signal plus noise amplitudes A_{j}.sup.(+) which are disposed within the second and third time windows, respectively.

The window sums are transmitted uphole from the NMR tool 13 of FIGS. 1 and 10 to the processing system or well logging truck computer 7a disposed on the wellbore surface as shown in FIG. 10. There are N_{w} window sums, where N_{w} is typically three to five in number. Therefore, since only N_{w} window sums are being transmitted uphole instead of 2J amplitudes (there being R_{j} amplitudes and X_{j} amplitudes, where j =1, 2, 3, . . , J), the telemetry requirements needed to transmit the N_{w} window sums uphole to the truck computer 7a, relative to the telemetry requirements needed to transmit the 2J amplitudes uphole, is significantly reduced.

Referring again to FIG. 13, the second part 7a4A of the signal processing method and apparatus of the present invention is illustrated.

In FIG. 13, recall that the RMS noise (SQRT psi) is output from the first part 46a1 of the signal processing method and apparatus purposes:

1. to compute "gamma", block 4A1--the computation of "gamma" is discussed in connection with equation (42) of the Detailed Description and is set forth in Appendix A of the Detailed Description of the Preferred Embodiment, entitled "An Algorithm for Optimal Selection of gamma"; in Appendix A, note that the best value of "gamma" can be found by finding the roots of equation (A.26) or of similarly derived equations; and

2. to compute standard deviations "sigma(phi_{nmr})", "sigma(phi_{f})", and "sigma(T_{2},log)", block 4A2--the standard deviation "sigma(phi_{nmr})" may be determined from equation (58) of the Detailed Description of the preferred embodiment set forth below; the standard deviation "sigma(phif)" may be determined from equation (61) of the Detailed Description; and the standard deviation "sigma(T_{2},log)" may be determined from equation (65) of the Detailed Description set forth below.

The parameter "gamma", computed in block 4A1, is used to construct and minimize the likelihood function (-ln L), block 4A3. The likelihood function (-ln L) is represented by the following equation, which is equation (42) of the Detailed Description set forth below: ##EQU3##

The spectral amplitudes (a_{1}) are determined by minimization of equation (42) subject to a positivity constraint, as indicated in the Detailed Description set forth below in connection with equation 42.

The spectral amplitudes (a_{1}) are used for two purposes: to compute log outputs "phi_{nmr} ", "phif", and T_{2},log; and to compute signal distributions P_{a} (log T_{2}) represented by color maps 2A1 of FIG. 16, to be discussed below.

To compute log outputs "phi_{nmr} ", "phi_{f} ", and T_{2},log, block 4A4, the log output "phi_{nmr} " is determined from equation (43), as follows: ##EQU4##

The log output "phi_{f} " is determined from equation (44), as follows: ##EQU5##

The log output T_{2},log is determined frcm equations (54) and (55), as follows: ##EQU6##

To compute signal distributions P_{a} (log T_{2}) which represent color maps 2A1 of FIG. 16, block 4A5, the computation of the signal distributions P_{a} (log T_{2}) is performed using equation (50) as follows: ##EQU7##

Referring to FIG. 16, a new output record medium 7a2A is illustrated. This new output record medium, a new log adapted to be given to a client for evaluation of the formation traversed by the wellbore, is generated in response to the receipt of the following information:

1. the log outputs "phi_{nmr} ", "phi_{f} ", and T_{2},1og of block 4A4;

2. the signal distributions for color maps P_{a} (log T_{2}) of block 4A5; and

3. the standard deviations "sigma(phi_{nmr})", "sigma(phi_{f})", and "sigma(T_{2},log)" of block 4A2.

The new output record medium 7a2A of FIG. 16 records the following new data or information presented in the form of logs as a function of depth in the wellbore (see element 2A2 in FIG. 16) and in the form of a color map (see element 2A1 of FIG. 16);

1. signal phase 2A3 from block alA;

2. RMS noise estimate 2A4 from block alC;

3. free fluid standard deviation 2A5 from block 4A2;

4. porosity standard deviation 2A6 from block 4A2;

5. free fluid porosity 2A7 from block 4A4;

6. total NMR porosity 2A8 from block 4A4; and

7. color Map 2A1.

The new output record medium 7a2A is given to a customer or client for the purpose of determining the presence or absence of underground deposits of hydrocarbons and the reservoir quality of the formation traversed by the wellbore 10 of FIG. 1.

A functional description of the operation of the signal processing method and apparatus in accordance with the present invention is set forth in the following paragraphs with reference to FIG. 16a and FIG. 13 of the drawings.

In FIGS. 16a and 13, the RF antennas 18 and/or 21 measure the precession of the protons in the pores 19a1 of the volume of investigation 19 of FIG. 3 and generate spin-echo receiver voltage pulses similar to the spin-echo receiver voltage pulses "echo 1", "echo 2", and "echo 3" illustrated in FIG. 9b of the drawings. Phase sensitive detection (PSD) circuits disposed within the electronics cartridge 24 integrate each of the spin echo receiver voltage pulses over a time interval, and the integrated signals are recorded as spin-echo inphase (R_{j}) and quadrature (X_{j}) amplitudes. The processor 46c of the downhole computer 46 in the electronics cartridge 24 disposed within the NMR tool 13 in the wellbore begins executing the first part 46a1 of the signal processing (software) method and apparatus of FIG. 13 stored within the memory 6a of the downhole computer 46. When the processor 46c completes the execution of the first part of the signal processing software 46a1 stored in memory 46a, the following data and information is determined:

1. the signal phase (theta) is estimated from the 2J inphase (R_{j}) and quadrature (X_{j}) amplitudes corresponding to J spin-echo receiver voltage pulses in the R and X channels using equation 9 as a function of the inphase and quadrature amplitudes set forth in the detailed description, block a1A of FIG. 13;

2. a spin-echo signal plus noise amplitude A_{j}.sup.(+) is determined for each inphase (R_{j}) amplitude, quadrature (X_{j}) amplitude, and signal phase (theta) using equation 22 in the Detailed Description; and the amplitude A_{j}.sup.(-) is also determined from inphase (R_{j}) amplitude, quadrature (X_{j}) amplitude, and signal phase (theta) using equation 23 in the Detailed Description, block alB of FIG. 13; furthermore, there are J inphase amplitudes (R_{j}) and J quadrature amplitudes (X_{j}); and, as a result, there are J spin-echo signal plus noise amplitudes A_{j}.sup.(+) and there are J amplitudes A_{j}.sup.(-) ;

3. the N_{w} window sums I_{m},m+1 are determined from the A_{j} .sup.(+) signal plus noise amplitudes using equation 35, where N_{w} is typically three to five in number, thereby reducing telemetry requirements for transmission of window sums uphole to the surface computer 7a, block a1D of FIG. 13; and

4. the RMS noise SQRT PSI is determined from the J amplitudes A_{j}.sup.(-) using the practical implementation of equation 31 or equation per se, block a1C of FIG. 13.

Two sets of data are transmitted uphole from the NMR tool 13 to the surface computer 7a: the N_{w} Window sums I_{m},m+1 and the RMS noise SQRT PSI.

The processor 7a3 of the surface computer 7a receives the N_{w} window sums I_{m},m+1 and the RMS noise SQRT PSI from the downhole computer 46 of the NMR tool 13 disposed downhole; in response, the processor 7a3 begins to execute the second part of the signal processing software 7a4A of FIG. 13 stored in memory 7a4 of the surface computer 7a. When the processor 7a3 completes execution of the second part of the signal processing software 7a4A, the following additional data and information is determined:

1. the standard deviations "sigma(phi_{nmr})", "sigma(phi_{f})", and "sigma(T_{2},log) " are determined, block 4A2 of FIG. 13, the standard deviation "sigma(phi_{nmr})" being determined from equation (58) of the Detailed Description, the standard deviation "sigma(phi_{f})" being determined from equation (61) of the Detailed Description, and the standard deviation "sigma(T_{2},log)" being determined from equation (65) of the Detailed Description of the Preferred Embodiment set forth below.

2. the parameter "gamma" is determined from the RMS noise SQRT PSI, block 4Al of FIG. 13, in the manner described in Appendix A of the Detailed Description and in connection with equation 42 in the Detailed Description;

3. once the parameter "gamma" is determined, a likelihood function (-ln L) is constructed and minimized, the likelihood function being represented by equation 42 of the Detailed Description, which equation is a function of the parameter "gamma", block 4A3 of FIG. 13;

4. the spectral amplitudes {a_{1} } are determined by minimization of the likelihood function (-1n L) of equation 42, and the spectral amplitudes {a_{1} } are used for two purposes: to compute log outputs "phi_{nmr} ", "phi_{f} ", and T_{2}, log, block 4A4 of FIG. 13, and to compute the signal distributions P_{a} (log T_{2}), which signal distributions are illustrated in the form of the color maps 2A1 of FIG. 16, block 4A5 of FIG. 13;

The processor 7a3 of FIG. 16a instructs the recorder 7a2 to generate the new output record medium 7a2A of FIG. 16 using the aforementioned recently determined standard deviations "sigma(phi_{nmr})", "sigma(phi_{f})", and "sigma(T_{2},log)"; the signal distributions P_{a} (log T_{2}); and the log outputs "phi_{nmr} ", "phi_{f} ", and T_{2},log, the new output record medium 7a2A of FIG. 16 displaying the following new information:

1. signal phase 2A3 from block a1A;

2. RMS noise estimate 2A4 from block a1C;

3. free fluid standard deviation 2A5 from block 4A2;

4. porosity standard deviation 2A6 from block 4A2;

5. free fluid porosity 2A7 from block 4A4;

6. total NMR porosity 2A8 from block 4A4; and

7. color map 2A1.

The development of Pulsed Nuclear Magnetic Resonance logging tools to acquire downhole spin-echo measurements in earth formations penetrated by boreholes is a new technology. The measurement principles and pulse sequences have been recently published.^{1},2 This new logging technology provides detailed formation evaluation information previously obtainable only from costly laboratory analysis of conventional core data. This information presently includes but is not limited to: (1) total NMR porosity (φ_{nmr}), (2) free-fluid porosity (φ_{f}) (i.e., part of total porosity which is movable), (3) bound fluid porosity (φ_{b}), (4) spin-spin (T_{2}) relaxation time distributions which are related to pore size distributions in sandstones, and (5) continuous permeability logs in sandstones. To extract this information from the measured spin-echoes requires a signal processing algorithm which is capable of providing an accurate and repeatable spectral decomposition of the measured data. The algorithm must be efficient and robust for real time processing of data as it is continuously acquired by a moving logging tool. This report describes a new algorithm which statisfies these conditions and provides PNMT logs which agree well with conventional core and other log data. Other approaches to spectral decomposition of NMR data have been reported^{3-6}. Kenyon, et al.^{5} used the algorithm reported by Gallegos and Smith^{3} to compute T_{1} distributions in rocks from laboratory NMR inversion recovery data. Latour, et al.^{6} used a computationally expensive singular value decomposition algorithm to compute relaxation time distributions in rocks from laboratory NMR data.

It is well-known that sedimentary rocks have a distribution of pore sizes. This results in NMR signals in rocks which decay with a distribution of relaxation times. Mathematically, the signal processing problem is to determine the distribution functions from the measured data (i.e., solve an inverse problem). The aforementioned formation evaluation parameters are computed from the distribution functions. Generally, there exists a joint distribution function, i.e., a function of both the longitudinal (T_{1}) and transverse (T_{2}) spin relaxation times, however, laboratory measurements have shown that these distributions in sedimentary rocks are correlated.^{6} The correlation is valid at NMR frequencies in the range of 2 MHz and is applicable to the borehole measurements described in this report.

It suffices to describe the pulse sequence and typical measurement parameters used in practice. The spin-echo measurement employed is the well-known Carr-Purcell-Meiboom-Gill (CPMG) sequence. This sequence measures the decay of the amplitude of the transverse magnetization following a 90° (+x) r.f. pulse which rotates the magnetization into a plane transverse to an applied static magnetic field (H_{0} z). The frequency of the r.f. pulses is equal to the median Larmor proton precession frequency. The nuclear spins contributing to the transverse magnetization prcess at different Larmor frequencies in the inhomogeneous d.c. magnetic field. The free-induction decay (FID) signal which is generated by the 90° (+x) pulse decays to zero within microseconds because of spin dephasing produced by the spread in Larmor frequencies. The FID signal occus too soon following the 90° (+x) to be measured by the PNMT electronics. At times t_{j} =(2j-1)t_{cp} following the 90° (+x) pulse, a set of 180° (y) r.f. pulses are applied which cause the transverse magnetization to refocus at times t_{j} =2jt.sub. cp for j=1,2, . . . , J producing J spin-echo signals. Note that the r.f. magnetic field of the 180° pulses is transverse to both H_{0} and the r.f. magnetic field of the 90° pulses. The spin-echoes are equally spaced in time with spacing Δ=2t_{cp}. The train of spin-echoes represents a signal whose decay contains contributions from all components of the intrinsic T_{2} -distribution. Here intrinsic refers to a distribution that includes effects of microscopic spin-spin interactions as observed in bulk liquids as well as surface relaxation from the confining pores. It is the latter effects that frequently dominate in reservoir rocks and provide the link between T_{2} -distributions and pore size distributions. The effects, on the spin-echo decay, of molecular diffusion in the static field gradient can be made negligble for the PNMT by making the Carr-Purcell time t_{cp} sufficiently short. The CPMG sequence described above can be improved by using phase alternated pairs (PAPS) of CPMGs to eliminate baseline shifts. Phase alternated pairs of CPMGs differ by shifting the phase of the 90° pulses 180° degrees. The PNMT PAPS pulse sequences can therefore be written succinctly in an obvious notation.

CPMG.sup.(±) :W-90°(±x)-(t_{cp}-180°(y)-t_{cp}-(echo)_{j}),_{j=1},2, . . . , J (1)

where W is a wait time that must precede each CPMG so that the longitudinal magnetization can recover prior to the initial 90° pulse. The PAPS pairs are combined by taking one-half their difference. This eliminates baseline offsets. Note each PAPS is constructed by signal averaging two CPMG sequences which reduces the rms noise by a factor √2.

Typical values of the pulse parameters are t_{cp} =100-500 μs, W=0.5-1.5 s, and J=100-1000 echoes.

The spin-echo amplitudes are obtained by hardware integration of the receiver voltages over J time windows each centered about t_{j} =jΔ for j=1,2, . . . , J. The PNMT tool uses phase sensitive detection to measure the in-phase (R_{j}) and quadrature (X_{j}) components of the signal-plus-noise amplitudes. As shown in the next section, the signal phase θ is estimated from the R_{j} and X_{j} amplitudes which are then combined to provide the signal-plus-noise amplitudes A_{j}.sup.(+). A phenomenological model for the signal-plus-noise amplitude of the j-th echo can be written in the form: ##EQU8## where P_{j} (T_{1},T_{2}) is the joint relaxation time distribution function and the integals are over the domain of the distribution function. As noted previously, experiments have shown that to within a good approximation, the joint distribution function in reservoir rocks has the form,

P_{J}(T_{1},T_{2})=P(T_{2})δ(T_{1}-ξT_{2}), (3)

where ξ˜1.5. That is, the T_{1} and T_{2} distributions are practically identical except for a constant scaling factor. Substitution of eq. (3) into (2) and performing the integration over T_{1} leads to ##EQU9## where for A_{j}.sup.(+) expressed in volts, P(T_{2})dT_{2} is the signal amplitude in volts contributed by spectral components with transverse relaxation times in the interval T_{2} to T_{2} +dT_{2}. The domain of the function P(T_{2}) is the closed interval [T_{min}, T_{max} ]. The function f(W,ξT_{2}) accounts for the incomplete recovery of the longitudinal magnetization during the wait time W and is defined by, ##EQU10##

The integrals in eqs. (2) and (4) are the amplitudes of the transverse magnetization observed at the j-th echo and N_{j}.sup.(+) is the thermal noise in the tool electronics.

Equation (4) is a Fredholm integral equation of the first kind for the distribution function P(T_{2}). The solution of eq. (4) for P(T_{2}) from the measured spin-echo signal-plus-noise amplitudes is the signal processing problem that must be confronted to obtain maximum information from the PNMT data. This type of problem represents an inverse problem of the type frequently encountered in remote sensing problems.^{7} The relaxation time distribution is the central quantity of interest since essentially all of the petrophysical quantities of interest can be computed from this function.

As noted previously, the amplitudes of the spin-echoes are integrated over time windows and the integrated signals are recorded as R_{j} and X_{j} time series channels or waveforms. Each time series or CPMG.sup.(+) is combined with its phase alternated pair CPMG(-). The PAPS pairs are accumulated as the tool moves and then averaged and output into depth bins. Thus, in each depth bin the data typically consists of several hundred echoes R_{j} and X_{j} for j=1,2, . . . , J where J is the total number of echoes collected. Prior to applying the signal processing algorithm, the data in each depth bin are pre-processed. First, an estimate θ of the signal phase is computed. Using θ the R_{j} and X_{j} data are combined into two random time series A_{j}.sup.(+) and A_{j}.sup.(-). The random variables A_{j}.sup.(+) have the statistical properties of the phase coherent signal-plus-noise and are used to estimate the relaxation time distribution (i.e., by solving a discretized version of eq. (4)). The random variables A_{j}.sup.(-) have the statistical properties of the noise and are used to estimate the rms noise, √ψ, on a single echo of the PAPs pairs in each depth bin. In each depth bin the number of PAPs pairs accumulated and averaged depends on the pulse sequence parameters and the logging speed.

The spin-echo in-phase and quadrature amplitudes can be written in the form,

R_{j}=S_{j}cosθ+N_{j}.sup.(c), (6)

X_{j}=S_{j}sinθ+N_{j}.sup.(s), (7)

where θ is the signal phase and S_{j} is the signal. N_{j}.sup.(s) are thermal noise voltage fluctuations in the R and X receiver channels, respectively. The thermal noise fluctuations are uncorrelated, zero mean Gaussian random variables with the following statistical properties:

<N_{j}.sup.(c) >=<N_{j}.sup.(s) >=0, (8a)

<N_{j}.sup.(c) N_{k}.sup.(s) >=0, (8b)

<N_{j}.sup.(c) N_{k}.sup.(c) >=<N_{j}.sup.(s) N_{k}.sup.(s) >=δ_{j},k ψ. (8c)

The angular brackets are used to denote statistical or ensemble averages and δ_{j},k is the Kronecker delta function. Note that thermal noise fluctuations are a time translationally invariant stochastic process. Therefore it follows from the ergodic therorem taht statistical averages can be replaced by time averages. In eq. (8c), ψ is the noise variance on a single echo.

An unbiased estimator θ of the signal phase is computed from the in-phase and quadrature amplitudes, ##EQU11##

To compute the mean and variance of θ, sum Eqs. (6) and (7) over all echoes, i.e., ##EQU12## from which it follows that, ##EQU13## where I have defined, ##EQU14##

Combining eqs. (12) and (13) one finds to lowest order in the noise fluctuations that, ##EQU15## from which is follows that ##EQU16## where in obtaining the above equation I have used the result, ##EQU17## which is valid for θ≃θ. It follows easily from eq. (17) and the statistical properties of the noise that,

<θ>=θ, (19)

so that θ is an unbiased estimator of the signal phase θ. One also finds using the noise properties that, ##EQU18##

Finally, it follows from eqs. (19) and (20) that the variance in the phase estimator θ is given by, ##EQU19##

It should be noted that the phase estimator θ is sensible provided that the signal-to-noise ratio is not too small. In practice, the signal phase is found to be relatively constant in zones with porosities greater than a few p.u. and exhibits random fluctuations in zones with no appreciable signal. It is a useful tool diagnostic since it should not vary with the formation and should be relatively constant in porous zones during a logging run.

Using θ, it is convenient to construct two random time series from the R_{j} and X_{j}. These are the signal-plus-noise amplitudes A_{j}.sup.(+) introduced in eq. (2) and A_{j}.sup.(-) which can be used to estimate ψ from the data. These random time series are defined by,

A_{j}.sup.(+) =R_{j}cosθ+X_{j}sinθ, (22)

and

A_{j}.sup.(-) =R_{j}sinθ-X_{j}cosθ. (23)

The rationale for introducing A_{j}.sup.(+) is that it has the statistical properties of a phase coherent signal. The above pair of equations has a simple vectorial interpretation, e.g. in eq. (22) the two terms on the right are the projections of the R and X components along the total signal whereas the terms in eq. (23) are projections perpendicular to the total signal. To calculate its expectation value first note that,

A_{j}.sup.(+) =S_{j}cos(θ-θ)+N_{j}.sup.(+) ≃S_{j}+N_{j}.sup.(+), (24)

where I have used eqs. (6), (7) and (22) and defined,

N_{j}.sup.(+) =N_{j}.sup.(c) cosθ+N_{j}.sup.(s) sinθ. (25)

I have dropped terms in eq. (24) of order (θ-θ)^{2} which are assumed to be negligble. The expectation value of eq. (24) is obtained from the statistical properties of the noise, i.e., eqs. (8).

<A_{j}.sup.(+) >≃S_{j}. (26)

One also finds that,

<<A_{j}.sup.(+))^{2}>≃S_{j}^{2}+ψ, (27)

from which it follows that the variance in A_{j}.sup.(+) is ψ. I shall make use of these results in the development of the algorithm for computing the spectral distrubtion function. Next it is shown that A_{j}.sup.(-) has the properties of the noise. Substituting eqs. (6), (7) into (23) one finds that,

A_{j}.sup.(-) =S_{j}sin(θ-θ)+N_{j}.sup.(-) ≃N_{j}.sup.(-), (28)

where

N_{j}.sup.(-) =N_{j}.sup.(c) sinθ-N_{j}.sup.(s) cosθ. (29)

Using the statistical properties of the noise one finds that,

<(A_{j}.sup.(-) >≃0, (30)

and,

<<A_{j}.sup.(-))^{2}>≃ψ, (31)

from which it follows that the variance in A_{j}.sup.(-) is ψ. Thus, as noted previously, the random time series for A_{j}.sup.(-) can be used to estimate the rms noise √ψ.

First, the signal S_{j} which is given by the integral in eq. (4) is discretized, i.e., ##EQU20## where it is assumed that there are N_{3} , components (i.e., basis functions) in the spectrum with relaxation times given by, ##EQU21## for l=1,2, . . . , N_{3}. Note that the T_{2},l are equally spaced on a logarithmic scale. In Eq. (32), I have defined the polarization functions, ##EQU22##

The signal processing problem therefore becomes the determination of the N_{3} spectral amplitudes α_{l} ≡P(T_{2},l)δ_{l} where δ_{l} =(T_{2},l+1 -T_{2},l-1)/2.

In this section, the data A_{j}.sup.(+) are reduced by introducing sums of echoes over time windows. As noted earlier, the different windows have different sensitivites to the various components of the spectrum. Consider N_{w} non-overlapping windows. Define the window sum, I_{m},m+1, over (N_{m+1} -N_{m} +δ_{m},1) echoes in the m-th window, i.e., ##EQU23## where ρ_{m} =(1-δ_{m},1) for m=1,2, . . . N.sub.ω and δ_{m},1 is the familiar Kronecker delta function (e.g., equals 1 for m=1 and 0 otherwise). Here (N_{m} +P_{m}) and N_{m+1} are the echo numbers (i.e., endpoints) of the m-th window. More explicitly, for the first window ##EQU24## where N_{1} and N_{2} are the echo numbers of the endpoints of the first window. For the second window, ##EQU25## and, in general, for the n-th window where n>1, ##EQU26##

Recalling eq. (32), one can write (35) in the form, ##EQU27## where I have defined window spectral sensitivity functions F_{m},m+1 given by geometric series, i.e., ##EQU28## and also defined the sums over noise, ##EQU29##

In each window the set of sensistivity functions defined above for l=1, . . . N, provide the relative sensitivity of each window to the different components in the spectrum.

Performing the summation in eq. (37), one finds that the sensitivity functions can be expressed in a form convenient for computation, i.e., ##EQU30## where x_{1} =Δ/T_{2},1. In practice, it has been found that for PNMT log data that usually 3 windows are sufficient (i.e., the log outputs are not altered by adding more windows). Likewise for continuous logging it has been found that N_{s} =8 is sufficient. This is discussed further in a section that examines the statistical uncertainties in the log outputs.

The statistical properties of the reduced set of random variable I_{m},m+1 are determined from the known statistical properties of the noise. One finds from eq. (36) that their expectation values are given by, ##EQU31## where the curly brackets on the right are used to indicate that the expectation value is a functional (i.e., a scalar whose value depends on a function) of the spectral amplitudes. The above expression has a simple physical interpretation, i.e., the expected value of the sum of the signal over the m-th window is the weighted sum of the sensitivity functions of the various components each weighted by its amplitude. The variances in the window sums are easily calculated. One finds that,

σ^{2}(I_{m},m+1)=ψ(N_{m+1}-N_{m}+δ_{m},1)≡ψσ_{m},m+1. (41)

Here ψ is the variance of the noise in a single echo which is the same for all echoes. The variance of the window sum in the above equation is simply the number of echoes in the window times the variance in a single echo. Note that this result is evident from elementary statistics since the variance in a sum of uncorrelated random variables is simply the sum of the variances. For non-overlapping windows the window sums over different windows are uncorrelated, i.e.,

<δI_{m},m+1 δI_{m},m+1 >=δ_{m},m ψσ_{m},m+1^{2}, (41a)

where from eq. (36), δI_{m},m+1 =N_{m},m+1.

The window sums defined in eq. (35) are independent Gaussian random variables with expectation values and variances given by eqs. (40) and (41), respectively. The logarithm of the likelihood function for the N.sub.ψ window sums is given by, ##EQU32## where the expected values <I_{m},m+1 >≡I_{m},m+1{a_{1} } were defined earlier and are functions of the spectral parameters which we want to estimate. The last term is a phenomenological regularization term introduced to prevent noise amplification artifacts. It is a penalty constraint that prevents the L_{2} norm of the amplitudes from becoming too large. The L_{2} regulatization is commonly employed but other criteria can be imposed. The spectral amplitudes are relatively insensitive to the dimensionless parameter δ. A value γ≃5 is usually used for PNMT log data. How to best choose γ is not a trivial question. The best fit to the data occurs if γ=0, however, the solution is not stable. It would be the best solution in the absence of noise. In the presence of noise, the squared residuals of the fit of the data in each window should be eaual to the variance in the window sums in eq. (41). This represents the best fit based on the statistics of the model. Increasing γ reduces the variance in the estimates but can lead to solutions that do not fit the data if γ is too large. An algorithm for a priori estimation of an "optimal" γ determined at each depth from the data is developed in Appendix A. The spectral amplitudes {a_{1} } are determined by minimization of eq. (42) subject to a positivity constraint. The estimation is extremely fast on a computer as only a few (e.g., 3) windows are needed. The computation time is essentially independent of the number of echoes.

The above algorithm leads to a tremendous data reduction since the spectrum is obtained from only a few random variables instead of hundreds or thousands of echoes. This huge data reduction has obvious potential benefits for efficient data acquisition and storage. A set (e.g., 10-100) of window sums can be rapidly computed downhole and transmitted uphole for processing. This set can be combined into gates for uphole processing. This leads to a substantial reduction in PNMT telemetry requirements which is important for commercial tools run in combination. Downhole preprocessing can be used to access data quality and flags established for sending all of the echoes uphole if necessary.

The spectral analysis described above leads immediately to logs of total, bound and free fluid porosities and mean transverse spin relaxation time. The total NMR porosity φ_{nmr} can be computed by integration of the distribution function P(T_{2}) defined in eq. (4), i.e., ##EQU33## where K_{tool} is a tool constant for converting volts to porosity. Similarly, the free-fluid porosity is given by, ##EQU34## where for the free fluid porosity (denoted by UBF on PNMT logs), the summation is over a subset of components, I=N_{c}, N_{c} +1, . . . N_{s} for which T_{2},1 ≧T_{c}. Centrifuge experiments by Straley, et al.^{8} have found that in many sandtones T_{c} ≃33 milliseconds. The bound-fluid porosity is simply given by φ_{bf} =φ-φ_{f}. The term Δφ is a small correction which accounts for the fact that the cut-off T_{c} does not lie on the endpoint of the free fluid integration interval. An analytic expression is derived in Appendix C for the Δφ correction. Note that this correction does not affect φ_{t}, i.e., only the partitioning of the free and bound fluid porosities.

The porosity distribution function P(T_{3}) is with respect to T_{2}. For displaying maps of porosity distributions versus relaxation time, it is useful to define a logarithmic distribution P_{a} (logT_{2}) since the relaxation times span several decades. In terms of the logarithmic distribution function, ##EQU35## where K_{tool} P_{a} (logT_{2})d(logT_{2}) is the porosity in the interval [logT_{2},logT_{2} +d(logT_{2})]. The distribution with respect to T_{2} and logT_{2} are related, i.e., ##EQU36## with c=(ln10)^{-1} as can be shown using eqs. (43) and (45). Using the discretization of P(T_{2}) introduced earlier, the discretized distribution with respect to the T_{2},1 defined in eq. (33) is given by, ##EQU37## with δ_{1} =(T_{2},1+1 -T_{2},1-1)/2. Using eq. (33), it follows from simple algebra that, ##EQU38## where I have defined, ##EQU39## Combining eqs. (46)-(48), one finds that, ##EQU40## Note that to within a constant factor, independent of l, the logarithmic distribution is proportional to the amplitude distribution {a_{1} }.

The main results concerning distributions are eqs. (47) and (50). In summary, to display the distribution P(T_{2}), use eq. (47) to compute P(T_{2},1) and plot it versus T_{2},1 on a linear scale. To display, the logarithmic distribution use eq. (50) to compute P_{a} (logT_{2},1) and plot it versus T_{2},1 on a logarithmic scale. More simply, plot the {a_{1} } distribution on a logarithmic T_{2},1 scale to obtain the logarithmic distributin to within a constant scale factor.

A mean relaxation time T_{2} can be defined for the distribution P(T_{2}). That is, ##EQU41## or in discretized form, ##EQU42##

Analagously, a logarithmic mean relaxation time T_{2},log can be defined. First, one defines a mean logarithm m for the P_{a} (logT_{2}) distribution, i.e., ##EQU43## or in discreted form, ##EQU44## Note that on a logarithmic scale the spacings, δ=log T_{2},l+1 -log T_{2},l ≡(N_{s} -1)^{-1} log[T_{max} /T_{min} ], are equal and therefore independent of l. The last equality in the above equation follows from the result in eq. (50). The logarithmic means relaxation time T_{2},log is obtained by exponentiation,

T_{2},log =10^{m}. (55)

The logarithmic means relaxation times T_{2},log and porosities φ_{nmr} are used as inputs into empirically derivatived equations to provide estimates of permeabilities.

The variances in the porosity estimators φ_{nmr}, φ_{f} and φ_{bf} are computed from the covariance matrix C_{l},k. The porosity estimators are obtained from estimates of the spectral amplitudes, i.e., ##EQU45## where the hat is used to differentiate the estimators from the true quantities (e.g., in eqs. (43)-(44) true porosities and spectral amplitudes are indicated). The variance σ^{2} (φ_{nmr}) is by definition,

σ^{2}(φ_{nmr})≡<φ_{nmr}^{2}>-(<φ_{nmr}>)^{2}, (57)

or explicity by, ##EQU46## where the parameter-parameter covariance matrix,

C_{l},k =<δa_{l}δa_{k}>, (59)

has been introduced. The angular brackets denote statistical averages and the fluctuations δa_{k} are defined by,

δa_{k}=a_{k}-<a_{k}>. (60)

Note the covariance matrix is symmetric and its diagonal elements are the variances in the amplitudes, i.e., σ^{2} (a_{l}). In general, the fluctuations in the amplitude estimates are correlated and therefore the off-diagonal elements of C_{l},k are non-zero. The derivation of the variances in the free and bound fluid porosities is analogous to the derivation of eq. (58). One finds, for example, that ##EQU47## where N_{c} is defined in eq. (44). In order to apply eps (58) and (61), one needs to compute the convariance matrix. An approximation for the covariance matrix will be derived in a subsequent section. First, however, an approximate calcuation of the variance in the logarithmic mean relaxation time is presented.

In this section an approximate formula for the variance T_{2},log is derived. Recalling eqs. (54) and (55) one finds that, ##EQU48## where the curly bracket on the left indicates that T_{2},log is a functional of the amplitude estimates a_{l}. It is to be understood that the summations on the right are over the whole spectrum, i.e., l=1, 2, . . . N_{s}. The constant c=(ln 10)^{-1}. Expand T_{2},log in a Taylor's series about the expectation values of the amplitude estimates, i.e., ##EQU49## Note that in eq. (63), it has been assumed that the fluctuations are sufficiently small that terms of order (δa_{k})^{2} can be neglected. The derivatives on the right are, of course, evaluated at <a_{k} >. To proceed, another small fluctuation approximation, i.e., <T_{2},log >≃T_{2},log (<a_{l} >) is employed to obtain ##EQU50## where, δT_{2},log ≡(T_{2},log -<T_{2},log >), and eqs. (59) and (63) have been used. The partial derivatives can be evaluated explicitly using eq. (62). One finds after some simple algebra that the standard deviation, ##EQU51## where I have defined the quantities, ##EQU52## where eq. (33) was used in obtaining the above result. Here δ is the T_{2},l spacing on a logarithmic scale (e.g., see the remarks following eq. (54)). In actual computations, the expectation values in the above equations are replaced by the maximum likelihood estimates obtained by the minimization of eq. (42). The calculation of the variance in T_{2} can be derived by similar manipulations but is not given here. The notion of a relaxation time becomes meaningless whenever the signal is dominated by the noise. This occurs in low porosity (e.g., for φ_{nmr} ≃1 p.u.) formations and leads to random fluctuations in T_{2},log usually occuring at long times since noise amplitude fluctuations are non-decaying. In these instances eq. (65) provides a useful criterion for "turning off" the T_{2},log log curve. A criterion which has proven useful is to disallow the log curve if the factor multiplying T_{2},log on the right side of eq. (65) exceeds unity.

An exact covariance matrix can be calculated for the algorithm. Using eqs. (A.1)-(A.3) and eq. (42), it is easy to prove that the parameter estimates are related to the "data" via the set of linear equations,

a_{l}=R_{l},m I_{m},m-1, (67)

for l=1, 2, . . . , N_{s} and where the Einstein summation convention of summing over repeated indices is used in this section. In the above equation, I have defined the N_{s} ×N_{w} matrix R,

R_{l},m =M_{l},k^{-1}Q_{k},m, (68)

where the N_{s} ×N_{s} matrix M is defined by, ##EQU53## and where the N_{s} ×N_{w} matrix Q is defined by, ##EQU54## where δ_{m},m+1 is defined in eq. (41). Using the definition of the parameter-parameter covariance matrix in eq. (59) and eqs. (41) and (41a) one finds that,

C_{l},k =R_{l},m ψσ_{m},m+1^{2}R_{m},k^{t}. (71)

The parameter-parameter convariance matrix in eq. (71) can be used with eqs. (58), (61) and with an analogous equation for σ^{2} (φ_{bf}) to study the standard deviations in φ_{t}, φ_{f} and φ_{bf} for various pulse and processing parameters. Some of these results are shown in FIGS. 17-20. The PNMT logs from repeat runs generally agree very well with the standard deviations computed from the covariance matrix. It should be noted, however, that log repeatability can be affected by many factors other than statistical fluctuations (e.g., hole conditions). Therefore the uncertainties computed using eq. (71) and displayed on the logs may in some cases be optimistic compared to the log uncertainties estimated from statistical analysis of repeat runs. A proof that the matrix M is positive definite and therefore has an in Appendix B.

The algorithm described above has been used to process log data from five Schlumberger clients wells logged to date. A flowchart illustrating the various steps in the implementation of the algorithm in shown in FIG. 13.

A detailed discussion of the field examples is beyond the scope of this report. Here, it will suffice to show a short section of processed continuous PNMT log data and the analysis of station data acquired at several depths within the interval. The log data shown was acquired with the PNMT tool moving at 150 ft/hour. FIG. 16 shows a section of log from a well in Texas. The section shown contains a non-hydrocarbon bearing thinly bedded sand/shale sequence. In Track 1, a color map of signal versus relaxation time (T_{2}) plotted on a logarithmic scale is shown. The magnitude of the signal is proportional to the intensity of the color and in this example, the T_{2} range is from 1 ms to 1500 ms. The red curve in Track 1 is the logarithmic means (T_{2},log) of the distribution. In Track 2, the rms noise estimate (√ψ) in volts, the estimated standard deviations in the total and free-fluid porosities, and the signal phase estimate (θ) in degrees are displayed. Note that θ is essentially constant, as expected, except at depths with low porosities (i.e., low signals) where the phase estimator is dominated by random noise. In Track 3, the total and free-fluid porosity estimates are displayed.

Several station stop measurements were made in the logged interval. At station stops, the tool acquires data for several minutes. The data is averaged to reduce the noise so that the quality of the station data is significantly better than the continuous data. The signal processing of the station data is the same as for the continous data.

In FIG. 21, the signal distribution (analogous to the color map shown in FIG. 16) versus T_{2} is shown for a station stop at 1100 ft. Recall that the area under the signal distribution curve is proportional to porosity. Comparing FIG. 21 with the continuous log observe that the total NMR porosity (φ_{nmr}), free-fluid porosity (φ_{f}), and logarithmic means relaxation time (T_{2},log) computed from the station stop data agree well with the continuous log values. The values shown were computed with a regularization parameter, γ=5.0, as was the continuous log. It should be noted, that it has been found that estimates of φ_{nmr}, φ_{f} and T_{2},log depend only weakly on the regulariation parameter. For example, changing γ by an order of magnitude usually results in porosity variations of less than ±0.5 pu. The detailed shape of the signal distribution, however, can be changed significantly in some cases by varying the regulation parameter. This is a consequence of the fact that there are infinitely many solutions which will fit the data. Decreasing γ reduces the fit error but can increase the norm of the solution vector.

In FIG. 21, the two signal distributions plotted correspond to two very different values of the regularization parameter. Note that there is practicially no difference in the two distributions. Both distributions fit the data to within the noise and have comparable norms. Therefore both solutions are mathematically acceptable.

In FIG. 22, results from a station stop at 1125 ft are displayed. The two signal distributions corresponding to values of γ differing by an order of magnitude are qualitatively similar. The distribution computed with γ=0.5 amplifies the two peak structure already apparent in the distribution computed using γ=5.0. In this example, the two peaks are probably due to signal contributions from two disparate pore distributions in the thinly bedded heterogeneous reservoir at 1125 ft as indicated on the FMS image (not shown here). In oil reservoirs, where the formation is relatively homogeneous over the length of the PNMT tool aperture, separate oil and water signal peaks can be identified in the signal distribution.

In FIG. 23, results from a station stop at 1148 ft are shown. Note that, the continuous log outputs agree will with those obtained by processing the station data. Note that, the reservoir quality at this depth is poor compared to that at the previous two stations as evidenced by almost all of the signal being associated with bound fluid porosity.

In FIGS. 24 and 25 the station stops at 1200 ft and 1229 ft are shown. These distributions, reveal relatively high permeability reservoir rock at these depths. The long relaxation times are indicative of large pore and, pore surfaces free of iron or other magnetic material.

The following references are incorporated into this specification by reference:

1. Sezginer, A., Kleinberg, R. L., Fukuhara, A., and Latour, L. L., "Very Rapid Measurement of Nuclear Magnetic Resonance Spin-Lattice Relaxation Time and Spin-Spin Relaxation Time," J. of Magnetic Resonance, v. 92, 504-527 (1992).

2. Kleinberg, R. L., Sezginer, A., Griffin, D. D., and Fukuhara, M., "Novel NMR Apparatus for Investigating an External Sample," J. of Magnetic Resonance, v. 97, 466-485 (1992).

3. Gallegos, D. P. and Smith, D. M., "A NMR Technique for the Analysis of Pore Structure: Determination of Continuous Pore Size Distributions," J. of Colloid and Interface Science, v. 122, No. 1, pp. 143-153, Mar., 1988.

4. Brown, R. J. S., Borgia, G. C., Fantazzini, P., and Mesini, E., "Problems In Identifying Multimodal Distributions Of Relaxation Times For NMR In Porous Media," Magnetic Resonance Imaging, Vol. 9, pp. 687-693 (1991).

5. Kenyon, W. E., Howard, J. J., Sezginer, A., Straley, C., Matteson, A., Horkowitz, R., and Erlich, R., "Pore-size Distribution and NMR in Microporous Cherty Sandstones," Trans. of the SPWLA of the 30th Ann. Logging Symp., Paper LL, Jun. 11-14, 1989.

6. Latour, L. L., Kleinberg, R. L., and A. Sezginer, "Nuclear Magnetic Resonance Properties of Rocks at Elevated Temperatures," J. of Colloid and Interface Science, v. 150, 535 (1992).

7. Twomey, S., "Introduction To The Mathematics of Inversion In Remote Sensing And Indirect Measurements," published by Elsevier Scientific Publishing Company, 1977.

8. Straley, C., Morriss, C. E., Kenyon, W. E., and Howard, J. J., "NMR in Partially Saturated Rocks: Laboratory Insights on Free Fluid Index and Comparison with Borehole Logs," Trans. of the SPWLA 32nd Ann. Logging Symp., Paper CC, Jun. 16-19, 1991.

9. Butler, J. P., Reeds, J. A. and Dawson, S. V., "Estimating Solutions of First Kind Integral Equations with Non-Negative Constraints and Optimal Smoothing," SIAM J. Numerical Anal., v. 18, No. 3, 381-397, 1981.

In this Appendix an algorithm for selection of an "optimal" regularization parameter γ_{opt} is derived. A similar algorithm was derived by Butler, Reeds and Dawson^{9} and has been used by Gallegose and Smith^{3} to select the regularization parameter. The selection criteion for choosing γ_{opt} is based on the notion that an optimal value of smoothing is obtained by minimizing the squared norm of the difference vector δa=a.sub.γ -a_{true}, where a.sub.γ is the regularized solution and a_{true} is the true distribution. Since a_{true} is, of course unknown, a compromise is made by replacing a_{true} by a hypothetical nose free solution a^{0} for which γ=0. Although, the aforementioned criterion can be rigorously stated, assumptions must be made that introduce elements of empiricism into the algorithm. These elements are also present in other algorithms^{9} and therefore the claims of optimally are somewhat subjective.

Maximum likelihood estimates of the N_{s} spectral amplitudes a_{l} are obtained by minimization of, -ln L, in eq. (42). The equations to be solved are ##EQU55## for l=1, 2, . . . N_{s}. Substituting eq. (42) into (A.1) leads to the linear system of equations, ##EQU56## where the symmetric matrix M_{l},k is defined in eq. (69) and the data vector ##EQU57## has been defined. The quantities f_{l}, I_{m},m+1 and F_{m},m+1 (T_{2},l) are defined in eqs. (34), (35) and (39), respectively, and

σ_{m},m+1^{2}=N_{m-1}-N_{m}+δ_{m},1, (A.4)

have been defined. The σ_{m},m+1^{2} are simply the number of echoes in the m-th window.

It is useful to write eq. (A.2) in an explicity matrix form,

M a.sub.γ =d, (A.5)

and to also write the companison equation,

M_{0}a.sup.(0) =d.sup.(0), (A.6)

where a.sub.γ is a vector whose N_{s} components are the spectral amplitudes for a regularization parameter γ, d is the data vector defined in eq. (A.3). The matrix M_{0} is defined by the matrix equation,

M=M_{0}+γI (A.7)

where IεR^{N}.sbsp.s.sup.×N.sbsp.s is the identity matrix. In (A.6), a.sup.(0) is the vector of spectral amplitudes corresponding to hypothetical noise free data and d.sup.(0) is the noise free data vector, i.e.,

d=d.sup.(0) +n, (A.8)

where the noise vector n is defined by, ##EQU58## where N_{m},m+1 is defined in eq. (38).

It follows from the theory of matrices that the real symmetric matrix M can be diagonalized by an orthogonal transformation, i.e.,

M=U D.sub.γ U^{t}, (A.10)

where U^{t} is the transpose of U where U is an N_{s} ×N_{s} matrix whose columns are an orthonormal set of eigenvectors of M, i.e., the vectors u_{j} satisfy the eigenvalue equation,

M u_{j}=λ_{j}(γ)u_{j}, (A.11)

and the othonormality conditions, ##EQU59## or the matrix equivalent,

U^{t}U=I. (A.12b)

where λ_{j} (γ) is the eigenvalue of M associated with eigenvector u_{j}. The N_{s} ×N_{s} matrix D.sub.γ in eq. (A.10) is a diagonal matrix with the eigenvalues λ_{j} (γ) on the diagonal. The above equations result from the orthonormality of the columns of the matrix M. Note that the operator M does not have a null space, i.e., it is positive definite for γ>0.

It can also be proven for orthogonal transformation square matrices like U that the rows are likewise orthonormal which leads to the equations, ##EQU60## or its matrix equivalent,

U U^{t}=I. (A.13b)

The operator M_{0} has a non-trivial null space because the rank (denoted by r) of M_{0}, i.e., the dimension of its non-null space is less than N_{s}. M_{0} has reduced rank because the data are not independent. This is typical of most inversion problems and mathematically the problem is underdetermined (more unknowns than measurements). In mathematical terms,

M_{0}u_{j}=λ_{j}(0)u_{j}, (A.14a)

for j=1,2, . . . , r where the eigenvalues can be ordered so that λ_{1} (0)>λ_{2} (0)>. . . >λ_{r} (0)>0. Also in the null space,

M_{0}u_{j}=0, (A.14b)

for j=r+1, . . . , N_{s}. The operator M_{0} can be diagonalized in its non-null space by the transformation,

M_{0}=U_{r}D_{0}U_{r}^{t}, (A.15)

where U_{r} is an N_{s} ×r matrix whose columns are the eigenvectors u_{j} where j=1, 2, . . . , r, and D_{0} is a diagonal matrix with eigenvalues λ_{j} (0) for j=1, 2, . . . , r as diagonal elements.

It follows from (A.7) that the eigenvalues λ_{j} (γ)of M are simply related to those of M_{0},

λ_{j}(γ)=λ_{j}(0)+γ. (A.16)

To proceed, one uses the transformation (A.10) in eq. (A.5) to write,

a.sub.γ =M^{-1}d=U D.sub.γ^{-1}U^{t}d, (A.17)

and similarly using eq. (A.15) and (A.6),

a.sup.(0) =M_{0}^{-1}d.sup.(0) =U_{r}D_{0}^{-1}U_{r}^{t}d.sup.(0). (A.18)

At this point, there are several ways to proceed which all lead to the same results. The most direct approach is to use the above equations, and write the solutions in eqs. (A.17) and (A.18) as eigenvector expansions, i.e., from (A.17) one finds ##EQU61## are the projections of the data onto the eigenvectors. Likewise from (A.18), ##EQU62##

In obtaining the above equation, I have used (A.8)and have defined the projections Q_{0},i =u_{i} ^{t} ·n of the noise onto the non-null space of M_{0}. The criterion for selecting γ is that the squared norm of the difference vector, a.sub.γ -a^{0}, be a minimum. Therefore, one is led to the minimization of the function F.sub.γ where

F.sub.γ =(a.sub.γ -a)^{t}·(.sub.γ -a^{0}), (A.22)

where (·) denotes the ordinary scalar product.

Combining eqs. (A.19)-(A.22) and using the orthonormality conditions one finds that ##EQU63## where C_{1} is a constant independent of γ, i.e., ##EQU64##

To proceed with the minimization of F.sub.γ, it is necessary to select a direction for the vector Q_{0}. The function is maximized (minimized) with respect to the noise by choosing Q_{0} parallel (anti-parallel) to Q. A conservative approach, also followed by Butler, Reeds and Dawson^{9} is to assume that the vectors are parallel. This assumption is not rigorously justifiable but errs on the side of over smoothing (selecting too large a γ) which is much less dangerous than under smoothing which results in a wildly oscillatory and meaningless inverse. Thus one is lead to write,

Q_{0}=s Q, (A.25),

where the scalar s is determined below.

Requiring that the derivative of F.sub.γ with respect to γ vanish, leads to a trancendental equation whose solution determines γ_{opt}, i.e., ##EQU65##

According to the criterion used in this Appendix, an optimal value of γ can be found by finding the non-negative roots of eq. (A.26). The equation is solved numerically, by using Newton's method. Note that, by inspection of (A.26) that for noise free data (i.e., s=0) that γ=0 as required. From (A.26) one can prove that the equation always has a unique non-negative solution for s<1 which is bounded in the interval,

γ_{l}<γ_{opt}<γ_{u}, (A.30)

where γ_{l} =sγ.sub.γ (0) and γ_{u} =sγ_{1} (0).

To complete this Appendix, it remains to determine the scalar s in eq. (A.25). To this end, note that ##EQU66##

Using, (A.19a) one finds that, ##EQU67## where the d_{i} are the components of the data vector defined in (A.3) and, ##EQU68##

Similarly, using (A.21) one finds that, ##EQU69## where the n_{i} are the components of the noise vector defined in (A.9) and where it has been assumed that in computing s one can replace the random variable Q_{0} ^{t} ·Q_{0} by its expectation value.

Finally, using (A.9) and the statistical properties of the noise one finds that, ##EQU70##

Recall from (A.7) that,

M_{l},k =M_{l},k.sup.(0) +γδ_{l},k, (B.1)

where from (67) and (A.4), ##EQU71##

Note that M_{l},k.sup.(0) can be written in the form, ##EQU72## where I have defined the matrix, ##EQU73##

For an arbitrary vector VεR^{N}.sbsp.3.sup.×1, note that

V^{t}B^{t}BV≡∥BV∥^{2}≧0,(B.5)

so that M.sup.(0) is positive semi-definite. More specifically, since V is arbitrary, one can choose it to be any eigenvector of M.sup.(0), e.g., u_{j} from which it follows that γ_{j} (0)≧0. For γ>0, it follows from (A.15) that γ_{j} (γ)>0 so that M is positive definite.

Recall from eq. (44) that the free-fluid porosity is defined by, ##EQU74## where K_{tool} is the tool constant.

This Appendix derives the correction Δφ in the above equation. This correction is needed because the bound-fluid porosity cut-off T_{c} generally does not lie on the leftmost endpoint of the free-fluid porosity integration interval. The correction is in practive almost negligble. It does not affect φ_{nmr}, only its partitioning into free and bound fluid porosity. It can be determined exactly by specifying the closed interval [T_{min}, T_{max} ] and the number of integration points N_{s} which determines a set of logarithmically spaced T_{2},l values for l=1, 2, . . . , N_{s}. The l-th integration element is a rectangular strip of height P(T_{2},l), width δ_{l} and area a_{l} ≡P(T_{2},l)δ_{l} where, ##EQU75##

The integer N_{c} in (C.1) is defined such that T_{2},1 ≧T_{c} for l=N_{c}, . . . , N_{s}. The endpoints of the rectangular strips are at the midpoints of adjacent values of T_{2},1. For example, let τ_{l} denote the midpoint (also the leftmost inegration point for the area element a_{l}). Then by definition, ##EQU76## Note that the widths (δ_{l}) of the rectangular strips can be written in terms of the τ_{1}, i.e.,

δ_{l}=τ_{l+1}-τ_{l}. (C.4)

This above described picture is shown schematically in FIG. 26 for the two rectangular integration elements corresponding to l=N_{c} -1 and l=N_{c}.

In general, there are two cases to consider: (a) τ≦T_{c} or (b) τ>T_{c}. Case (a) is illustrated in FIG. 26. In case (a), the correction Δφ≦0, so that the correction subtracts porosity from the summation in (C.1) which includes a small amount of bound fluid porosity. The porosity correction is the sand shaded area in FIG. 26 multiplied by the tool constant. This correction increases the bound fluid porosity by an equal amount.

A simple calculation shows that for case (a): ##EQU77## and for case (b) one finds that, ##EQU78##

The invention being thus described, it will be obvious that the same may be varied in many ways. Such variations are not to be regarded as a departure from the spirit and scope of the invention, and all such modifications as would be obvious to one skilled in the art are intended to be included within the scope of the following claims.

Patentzitate

Zitiertes Patent | Eingetragen | Veröffentlichungsdatum | Antragsteller | Titel |
---|---|---|---|---|

US3617867 * | 17. Juli 1953 | 2. Nov. 1971 | Texaco Inc | Method and apparatus for well logging utilizing resonance |

US4350955 * | 10. Okt. 1980 | 21. Sept. 1982 | The United States Of America As Represented By The United States Department Of Energy | Magnetic resonance apparatus |

US4389613 * | 15. Apr. 1981 | 21. Juni 1983 | Chevron Research Company | Computer-controlled, portable pulsed NMR instrument and method of use |

US4408161 * | 15. Apr. 1981 | 4. Okt. 1983 | Chevron Research Company | Computer-controlled, portable spin echo NMR instrument and method of use |

US4412178 * | 15. Apr. 1981 | 25. Okt. 1983 | Chevron Research Company | Computer-controlled, portable pulsed NMR instrument and method of use wherein a DC magnetic field gradient is artificially introduced |

US4412179 * | 15. Apr. 1981 | 25. Okt. 1983 | Chevron Research Company | Computer-controlled, portable pulsed NMR instrument and method of use wherein the times of RF interrogation are distributed over at least a cycle at the nuclear magnetization precessional frequency |

US4480227 * | 27. Juni 1983 | 30. Okt. 1984 | Chevron Research Company | Portable pulsed NMR instrument and method of use |

US4710713 * | 11. März 1986 | 1. Dez. 1987 | Numar Corporation | Nuclear magnetic resonance sensing apparatus and techniques |

US4714881 * | 15. Juli 1986 | 22. Dez. 1987 | Mobil Oil Corporation | Nuclear magnetic resonance borehole logging tool |

US4717876 * | 13. Aug. 1986 | 5. Jan. 1988 | Numar | NMR magnet system for well logging |

US4717877 * | 25. Sept. 1986 | 5. Jan. 1988 | Numar Corporation | Nuclear magnetic resonance sensing apparatus and techniques |

US4717878 * | 26. Sept. 1986 | 5. Jan. 1988 | Numar Corporation | Nuclear magnetic resonance sensing apparatus and techniques |

US4933638 * | 19. Juni 1989 | 12. Juni 1990 | Schlumber Technology Corp. | Borehole measurement of NMR characteristics of earth formations, and interpretations thereof |

US5023551 * | 19. Dez. 1989 | 11. Juni 1991 | Schlumberger-Doll Research | Nuclear magnetic resonance pulse sequences for use with borehole logging tools |

US5055787 * | 5. Dez. 1989 | 8. Okt. 1991 | Schlumberger Technology Corporation | Borehole measurement of NMR characteristics of earth formations |

US5055788 * | 5. Dez. 1989 | 8. Okt. 1991 | Schlumberger Technology Corporation | Borehole measurement of NMR characteristics of earth formations |

Nichtpatentzitate

Referenz | ||
---|---|---|

1 | Brown, R. J. S., Borgia, G. C., Fantazzini, P., and Mesini, E., "Problems in Identifying Multimodal Distributions of Relaxation Times for NMR in Porous Media," Magnetic Resonance Imaging, vol. 9, pp. 687-693 (1991). | |

2 | * | Brown, R. J. S., Borgia, G. C., Fantazzini, P., and Mesini, E., Problems in Identifying Multimodal Distributions of Relaxation Times for NMR in Porous Media, Magnetic Resonance Imaging, vol. 9, pp. 687 693 (1991). |

3 | Butler, J. P., Reeds, J. A. and Dawson, S. V., "Estimating Solutions of First Kind Integral Equations with NON-Negative Constraints and Optimal Smoothing," SIAM J. Numerical Anal., vol. 18, No. 3, 381-397, 1981. | |

4 | * | Butler, J. P., Reeds, J. A. and Dawson, S. V., Estimating Solutions of First Kind Integral Equations with NON Negative Constraints and Optimal Smoothing, SIAM J. Numerical Anal., vol. 18, No. 3, 381 397, 1981. |

5 | Gallegos, D. P. and Smith, D. M., "NMR Technique for the Analysis of Pore Structure: Determination of Continuous Pore Size Distributions," J. of Colloid and Interface Science, vol. 122, No. 1, pp. 143-153, Mar. 1988. | |

6 | * | Gallegos, D. P. and Smith, D. M., NMR Technique for the Analysis of Pore Structure: Determination of Continuous Pore Size Distributions, J. of Colloid and Interface Science, vol. 122, No. 1, pp. 143 153, Mar. 1988. |

7 | Kenyon, W. E., Howard, J. J., Sezginer, A., Straley, C., Matteson, A., Horkowitz, R., and Erlich, R., "Pore-size Distribution and NMR in Microporous Cherty Sandstones", Trans. of the SPWLA of the 30th Ann. Logging Symp., Paper LL, Jun. 11-14, 1989. | |

8 | * | Kenyon, W. E., Howard, J. J., Sezginer, A., Straley, C., Matteson, A., Horkowitz, R., and Erlich, R., Pore size Distribution and NMR in Microporous Cherty Sandstones , Trans. of the SPWLA of the 30th Ann. Logging Symp., Paper LL, Jun. 11 14, 1989. |

9 | Kleinberg, R. L., Sezginer, A., Griffin, D. D., and Fukuhara, M., "Novel NMR Apparatus for Investigating and External Sample," J. of Magnetic Resonance, vol. 97, 466-485 (1992). | |

10 | * | Kleinberg, R. L., Sezginer, A., Griffin, D. D., and Fukuhara, M., Novel NMR Apparatus for Investigating and External Sample, J. of Magnetic Resonance, vol. 97, 466 485 (1992). |

11 | Latour, L. L., Kleinberg, R. L., and A. Sezginer, "Nuclear Magnetic Resonance Properties of Rocks at Elevated Temperatures," J. of Colloid and Interface Science, vol. 150, 535 (1992). | |

12 | * | Latour, L. L., Kleinberg, R. L., and A. Sezginer, Nuclear Magnetic Resonance Properties of Rocks at Elevated Temperatures, J. of Colloid and Interface Science, vol. 150, 535 (1992). |

13 | M. N. Miller, Z. Paltiel, M. E. Gillen., J. Granot, and J. C. Bouton, NUMAR Corp., Article: "Spin Echo Magnetic Resonance Logging Porosity and Freed Fluid Indes Determination" SPE 20561, 321-332. | |

14 | * | M. N. Miller, Z. Paltiel, M. E. Gillen., J. Granot, and J. C. Bouton, NUMAR Corp., Article: Spin Echo Magnetic Resonance Logging Porosity and Freed Fluid Indes Determination SPE 20561, 321 332. |

15 | Ram M. Narayanan, Article: "Data Compression in Remote Sensing Applications", IEEE Geoscience and Remote Sensing Society Newsletter, Sep. 1992. | |

16 | * | Ram M. Narayanan, Article: Data Compression in Remote Sensing Applications , IEEE Geoscience and Remote Sensing Society Newsletter, Sep. 1992. |

17 | Sezginer, A., Kleinberg, R. L., Fukuhara, A., and Latour, L. L., "Very Rapid Measurement of Nuclear Magnetic Resonance Spin-Lattice Relaxation Time and Spin-Spin Relaxation Time," J. of Magnetic Resonance, vol. 92, 504-527 (1992). | |

18 | * | Sezginer, A., Kleinberg, R. L., Fukuhara, A., and Latour, L. L., Very Rapid Measurement of Nuclear Magnetic Resonance Spin Lattice Relaxation Time and Spin Spin Relaxation Time, J. of Magnetic Resonance, vol. 92, 504 527 (1992). |

19 | Straley, C., Morriss, C. E., Kenyon, W. E., and Howard, J. J., "NMR in Partially Saturated Rocks: Laboratory Insights on Free Fluid Index and Comparison with Borehole Logs," Trans. of the SPWLA 32nd Ann. Logging Symp., Paper CC, Jun. 16-19, 1991. | |

20 | * | Straley, C., Morriss, C. E., Kenyon, W. E., and Howard, J. J., NMR in Partially Saturated Rocks: Laboratory Insights on Free Fluid Index and Comparison with Borehole Logs, Trans. of the SPWLA 32nd Ann. Logging Symp., Paper CC, Jun. 16 19, 1991. |

21 | Twomey, S., "Introduction to the Mathematics of Inversion in Remote Sensing and Indirect Measurements," published by Elsevier Scientific Publishing Company, 1977. | |

22 | * | Twomey, S., Introduction to the Mathematics of Inversion in Remote Sensing and Indirect Measurements, published by Elsevier Scientific Publishing Company, 1977. |

Referenziert von

Zitiert von Patent | Eingetragen | Veröffentlichungsdatum | Antragsteller | Titel |
---|---|---|---|---|

US5381092 * | 28. Sept. 1993 | 10. Jan. 1995 | Schlumberger Technology Corporation | Method and apparatus for compressing data produced from a well tool in a wellbore prior to transmitting the compressed data uphole to a surface apparatus |

US5442269 * | 10. Sept. 1993 | 15. Aug. 1995 | Fujitsu Limited | Robot control system |

US5486762 * | 15. Aug. 1994 | 23. Jan. 1996 | Schlumberger Technology Corp. | Apparatus including multi-wait time pulsed NMR logging method for determining accurate T2-distributions and accurate T1/T2 ratios and generating a more accurate output record using the updated T2-distributions and T1/T2 ratios |

US5497087 * | 20. Okt. 1994 | 5. März 1996 | Shell Oil Company | NMR logging of natural gas reservoirs |

US5519668 * | 26. Mai 1994 | 21. Mai 1996 | Schlumberger Technology Corporation | Methods and devices for real-time formation imaging through measurement while drilling telemetry |

US5596274 * | 31. Okt. 1994 | 21. Jan. 1997 | Schlumberger Technology Corporation | Determining bound and unbound fluid volumes using nuclear magnetic resonance pulse sequences |

US5644231 * | 4. März 1996 | 1. Juli 1997 | Schlumberger Technology Corporation | High pressure magnetic antenna assembly |

US5680043 * | 23. März 1995 | 21. Okt. 1997 | Schlumberger Technology Corporation | Nuclear magnetic resonance technique for determining gas effect with borehole logging tools |

US5796252 * | 15. Jan. 1997 | 18. Aug. 1998 | Schlumberger Technology Corporation | Nuclear magnetic resonance borehole logging apparatus and method for ascertaining a volume of hydrocarbons independent of a diffusion coefficient |

US6026560 * | 24. Juni 1997 | 22. Febr. 2000 | Schlumberger Technology Corporation | High pressure magnet assembly |

US6040696 * | 16. Sept. 1997 | 21. März 2000 | Schlumberger Technology Corporation | Method for estimating pore structure in carbonates from NMR measurements |

US6084408 * | 13. Febr. 1998 | 4. Juli 2000 | Western Atlas International, Inc. | Methods for acquisition and processing of nuclear magnetic resonance signals for determining fluid properties in petroleum reservoirs having more than one fluid phase |

US6094048 * | 18. Dez. 1997 | 25. Juli 2000 | Shell Oil Company | NMR logging of natural gas reservoirs |

US6097184 * | 31. Dez. 1997 | 1. Aug. 2000 | Schlumberger Technology Corporation | Nuclear magnetic resonance well logging to determine gas-filled porosity and oil-filled porosity of earth formations without a constant static magnetic field gradient |

US6140817 * | 26. Mai 1998 | 31. Okt. 2000 | Schlumberger Technology Corporation | Magnetic resonance well logging method and apparatus |

US6147489 * | 12. Juni 1997 | 14. Nov. 2000 | Schlumberger Technology Corporation | Method and apparatus for measuring total nuclear magnetic resonance porosity |

US6166540 * | 30. Juni 1997 | 26. Dez. 2000 | Wollin Ventures, Inc. | Method of resistivity well logging utilizing nuclear magnetic resonance |

US6177794 | 13. Mai 1998 | 23. Jan. 2001 | The Regents Of The University Of California | Use of earth field spin echo NMR to search for liquid minerals |

US6229308 | 29. Okt. 1999 | 8. Mai 2001 | Schlumberger Technology Corporation | Formation evaluation using magnetic resonance logging measurements |

US6255818 | 4. Aug. 1999 | 3. Juli 2001 | Schlumberger Technology Corporation | Method and apparatus for performing magnetic resonance measurements |

US6255819 | 25. Okt. 1999 | 3. Juli 2001 | Halliburton Energy Services, Inc. | System and method for geologically-enhanced magnetic resonance imaging logs |

US6342784 | 8. Mai 2000 | 29. Jan. 2002 | Wollin Ventures, Inc. | Method for resistivity well logging utilizing nuclear magnetic resonance |

US6366087 | 28. Okt. 1999 | 2. Apr. 2002 | George Richard Coates | NMR logging apparatus and methods for fluid typing |

US6388441 | 18. Okt. 2000 | 14. Mai 2002 | Baker Hughes Incorporated | Method for processing NMR data without phase-alternating-pair (PAP) averaging |

US6437564 * | 1. Mai 2001 | 20. Aug. 2002 | Baker Hughes Incorporated | Estimate of transversal motion of the NMR tool during logging |

US6466013 * | 18. Apr. 2000 | 15. Okt. 2002 | Baker Hughes Incorporated | Nuclear magnetic resonance measurements in well logging using an optimized rephasing pulse sequence |

US6512371 | 5. Juni 2001 | 28. Jan. 2003 | Halliburton Energy Services, Inc. | System and method for determining oil, water and gas saturations for low-field gradient NMR logging tools |

US6518756 | 14. Juni 2001 | 11. Febr. 2003 | Halliburton Energy Services, Inc. | Systems and methods for determining motion tool parameters in borehole logging |

US6522138 | 1. März 2001 | 18. Febr. 2003 | Schlumberger Technology Corporation | Resolution enhancement for sequential phase alternated pair nuclear magnetic resonance measurements |

US6525534 | 15. Juni 2001 | 25. Febr. 2003 | Halliburton Energy Services, Inc. | System and methods for NMR signal processing without phase alternated pair stacking |

US6531868 | 5. Juni 2001 | 11. März 2003 | Halliburton Energy Services, Inc. | System and methods for formation evaluation while drilling |

US6541969 | 14. Dez. 2000 | 1. Apr. 2003 | Halliburton Energy Services, Inc. | Method and apparatus for improving the vertical resolution of NMR logs |

US6545471 | 11. Dez. 2001 | 8. Apr. 2003 | Wollin Ventures, Inc. | Method for resistivity well logging utilizing nuclear magnetic resonance |

US6577125 | 18. Dez. 2000 | 10. Juni 2003 | Halliburton Energy Services, Inc. | Temperature compensated magnetic field apparatus for NMR measurements |

US6661226 | 10. Aug. 2000 | 9. Dez. 2003 | Halliburton Energy Services, Inc. | NMR apparatus and methods for measuring volumes of hydrocarbon gas and oil |

US6686738 | 4. Febr. 2002 | 3. Febr. 2004 | Baker Hughes Incorporated | Method for determining decay characteristics of multi-component downhole decay data |

US6766252 | 24. Jan. 2002 | 20. Juli 2004 | Halliburton Energy Services, Inc. | High resolution dispersion estimation in acoustic well logging |

US6825658 | 2. Apr. 2002 | 30. Nov. 2004 | George Richard Coates | NMR logging apparatus and methods for fluid typing |

US6833698 * | 15. Mai 2002 | 21. Dez. 2004 | Chevrontexaco U.S.A. Inc. | Methods of decoupling diffusion effects from relaxation times to determine properties of porous media containing fluids |

US6838875 | 10. Mai 2002 | 4. Jan. 2005 | Schlumberger Technology Corporation | Processing NMR data in the presence of coherent ringing |

US6856132 | 8. Nov. 2002 | 15. Febr. 2005 | Shell Oil Company | Method and apparatus for subterranean formation flow imaging |

US6859032 | 18. Dez. 2001 | 22. Febr. 2005 | Schlumberger Technology Corporation | Method for determining molecular properties of hydrocarbon mixtures from NMR data |

US6937014 | 24. März 2003 | 30. Aug. 2005 | Chevron U.S.A. Inc. | Method for obtaining multi-dimensional proton density distributions from a system of nuclear spins |

US6952096 | 5. Sept. 2003 | 4. Okt. 2005 | Schlumberger Technology Corporation | Method and apparatus for determining speed and properties of flowing fluids using NMR measurements |

US6956371 | 28. Jan. 2003 | 18. Okt. 2005 | Halliburton Energy Services, Inc. | Method and apparatus for detecting diffusion sensitive phases with estimation of residual error in NMR logs |

US6956372 | 11. Sept. 2003 | 18. Okt. 2005 | Halliburton Energy Services, Inc. | System and method for NMR logging with helical polarization |

US7015694 | 19. Aug. 2002 | 21. März 2006 | Baker Hughes Incorporated | NMR apparatus and method for stochastic pulsing of earth formations |

US7026814 | 19. Dez. 2003 | 11. Apr. 2006 | Schlumberger Technology Corporation | Tuning of nuclear magnetic resonance logging tools |

US7053611 | 4. Juni 2004 | 30. Mai 2006 | Schlumberger Technology Corporation | Method and apparatus for using pulsed field gradient NMR measurements to determine fluid properties in a fluid sampling well logging tool |

US7091719 | 30. Apr. 2004 | 15. Aug. 2006 | Schlumberger Technology Corporation | Method for determining properties of formation fluids |

US7095812 * | 24. Juni 2002 | 22. Aug. 2006 | Agere Systems Inc. | Reduced complexity receiver for space-time- bit-interleaved coded modulation |

US7135862 | 13. März 2001 | 14. Nov. 2006 | Halliburton Energy Services, Inc | NMR logging using time-domain averaging |

US7298142 | 27. Juni 2005 | 20. Nov. 2007 | Baker Hughes Incorporated | Method and apparatus for reservoir fluid characterization in nuclear magnetic resonance logging |

US7317990 | 10. Okt. 2005 | 8. Jan. 2008 | Schlumberger Technology Corporation | Distributed processing system for subsurface operations |

US7408150 | 25. Juni 2007 | 5. Aug. 2008 | Schlumberger Technology Corporation | Well logging method for determining formation characteristics using pulsed neutron capture measurements |

US7414405 | 2. Aug. 2005 | 19. Aug. 2008 | Pathfinder Energy Services, Inc. | Measurement tool for obtaining tool face on a rotating drill collar |

US7436184 | 15. März 2005 | 14. Okt. 2008 | Pathfinder Energy Services, Inc. | Well logging apparatus for obtaining azimuthally sensitive formation resistivity measurements |

US7436185 | 27. Juni 2005 | 14. Okt. 2008 | Schlumberger Technology Corporation | Highly integrated logging tool |

US7463027 | 30. Apr. 2004 | 9. Dez. 2008 | Halliburton Energy Services, Inc. | Systems and methods for deep-looking NMR logging |

US7501818 | 29. März 2007 | 10. März 2009 | Halliburton Energy Services, Inc. | System and methods for T1-based logging |

US7511487 | 27. Febr. 2007 | 31. März 2009 | Schlumberger Technology Corporation | Logging method for determining characteristic of fluid in a downhole measurement region |

US7557582 | 24. Juli 2008 | 7. Juli 2009 | Smith International Inc | Logging while drilling tool for obtaining azimuthally sensitive formation resistivity measurements |

US7558675 | 25. Juli 2007 | 7. Juli 2009 | Smith International, Inc. | Probablistic imaging with azimuthally sensitive MWD/LWD sensors |

US7564240 * | 19. Juni 2007 | 21. Juli 2009 | Schlumberger Technology Corporation | Method and apparatus for measuring free induction decay signal and its application to composition analysis |

US7733086 | 30. Okt. 2008 | 8. Juni 2010 | Halliburton Energy Services, Inc. | Systems and methods for deep-looking NMR logging |

US7755354 | 4. Nov. 2008 | 13. Juli 2010 | Halliburton Energy Services, Inc. | System and methods for T1-based logging |

US7821260 * | 12. Febr. 2008 | 26. Okt. 2010 | Baker Hughes Incorporated | NMR echo train compression using only NMR signal matrix multiplication to provide a lower transmission bit parametric representation from which estimate values of earth formation properties are obtained |

US8004279 * | 21. Mai 2009 | 23. Aug. 2011 | Baker Hughes Incorporated | Real-time NMR distribution while drilling |

US8022698 * | 31. Dez. 2008 | 20. Sept. 2011 | Baker Hughes Incorporated | Joint compression of multiple echo trains using principal component analysis and independent component analysis |

US8600115 | 10. Juni 2011 | 3. Dez. 2013 | Schlumberger Technology Corporation | Borehole image reconstruction using inversion and tool spatial sensitivity functions |

US8686723 * | 22. März 2010 | 1. Apr. 2014 | Schlumberger Technology Corporation | Determining the larmor frequency for NMR tools |

US8836328 * | 1. Febr. 2011 | 16. Sept. 2014 | Baker Hughes Incorporated | Acoustic excitation with NMR pulse |

US9052409 | 11. Juli 2008 | 9. Juni 2015 | Schlumberger Technology Corporation | Monte Carlo method for laplace inversion of NMR data |

US9201158 * | 24. Jan. 2012 | 1. Dez. 2015 | Schlumberger Technology Corporation | Estimating and displaying molecular size information of a substance |

US9658360 | 2. Dez. 2011 | 23. Mai 2017 | Schlumberger Technology Corporation | High resolution LWD imaging |

US20020163334 * | 13. März 2001 | 7. Nov. 2002 | Teruhiko Hagiwara | NMR logging using time-domain averaging |

US20030016012 * | 2. Apr. 2002 | 23. Jan. 2003 | Coates George Richard | NMR logging apparatus and methods for fluid typing |

US20030092311 * | 12. Nov. 2002 | 15. Mai 2003 | France Telecom | Comb and a method for making a branch connection to preexisting cabling |

US20030128032 * | 18. Dez. 2001 | 10. Juli 2003 | Heaton Nicholas J. | Method for determining molecular properties of hydrocarbon mixtures from NMR data |

US20030214287 * | 15. Mai 2002 | 20. Nov. 2003 | Boqin Sun | Methods of decoupling diffusion effects from relaxation times to determine properties of porous media containing fluids and multi-dimensional representation of those properties |

US20040001564 * | 24. Juni 2002 | 1. Jan. 2004 | Albert Chan | Reduced complexity receiver for space-time- bit-interleaved coded modulation |

US20040008027 * | 28. Jan. 2003 | 15. Jan. 2004 | Manfred Prammer | Method and apparatus for detecting diffusion sensitive phases with estimation of residual error in NMR logs |

US20040032258 * | 19. Aug. 2002 | 19. Febr. 2004 | Baker Hughes Incorporated | Stochastic NMR for downhole use |

US20040056658 * | 11. Sept. 2003 | 25. März 2004 | Peter Masak | NMR tool with helical polarization |

US20040189296 * | 24. März 2003 | 30. Sept. 2004 | Chevron U.S.A. Inc. | Method for obtaining multi-dimensional proton density distributions from a system of nuclear spins |

US20050030021 * | 30. Apr. 2004 | 10. Febr. 2005 | Prammer Manfred G. | Systems and methods for NMR logging |

US20050140368 * | 5. Sept. 2003 | 30. Juni 2005 | Schlumberger Technology Corporation | [method and apparatus for determining speed and properties of flowing fluids using nmr measurements] |

US20050156592 * | 19. Dez. 2003 | 21. Juli 2005 | Ernesto Bordon | Tuning of nuclear magnetic resonance logging tools |

US20050242807 * | 30. Apr. 2004 | 3. Nov. 2005 | Robert Freedman | Method for determining properties of formation fluids |

US20050270023 * | 4. Juni 2004 | 8. Dez. 2005 | Robert Freedman | Method and apparatus for using pulsed field gradient NMR measurements to determine fluid properties in a fluid sampling well logging tool |

US20060089816 * | 10. Okt. 2005 | 27. Apr. 2006 | Schlumberger Technology Corporation | Distributed processing system for subsurface operations |

US20060208738 * | 15. März 2005 | 21. Sept. 2006 | Pathfinder Energy Services, Inc. | Well logging apparatus for obtaining azimuthally sensitive formation resistivity measurements |

US20060290350 * | 27. Juni 2005 | 28. Dez. 2006 | Hursan Gabor G | Method and apparatus for reservoir fluid characterization in nuclear magnetic resonance logging |

US20060290353 * | 27. Juni 2005 | 28. Dez. 2006 | Schlumberger Technology Corporation | Pad assembly for logging tool |

US20060290354 * | 27. Juni 2005 | 28. Dez. 2006 | Schlumberger Technology Corporation | Highly integrated logging tool |

US20070030007 * | 2. Aug. 2005 | 8. Febr. 2007 | Pathfinder Energy Services, Inc. | Measurement tool for obtaining tool face on a rotating drill collar |

US20070241750 * | 29. März 2007 | 18. Okt. 2007 | Ridvan Akkurt | System and methods for T1-based logging |

US20080036457 * | 28. Aug. 2007 | 14. Febr. 2008 | Baker Hughes Incorporated | NMR Echo Train Compression |

US20080183390 * | 12. Febr. 2008 | 31. Juli 2008 | Baker Hughes Incorporated | NMR Echo Train Compression |

US20080204013 * | 27. Febr. 2007 | 28. Aug. 2008 | Schlumberger Technology Corporation | Logging method for determining characteristic of fluid in a downhole measurement region |

US20080284440 * | 24. Juli 2008 | 20. Nov. 2008 | Pathfinder Energy Services, Inc. | Logging while drilling tool for obtaining azimuthally sensitive formation resistivity measurements |

US20080315873 * | 19. Juni 2007 | 25. Dez. 2008 | Schlumberger Technology Corporation | Method and Apparatus for Measuring Free Induction Decay Signal and Its Application to Composition Analysis |

US20090030616 * | 25. Juli 2007 | 29. Jan. 2009 | Pathfinder Energy Services, Inc. | Probablistic imaging with azimuthally sensitive MWD/LWD sensors |

US20090072825 * | 30. Okt. 2008 | 19. März 2009 | Prammer Manfred G | Systems and methods for deep-looking nmr logging |

US20090174402 * | 31. Dez. 2008 | 9. Juli 2009 | Baker Hughes Incorporated | Joint Compression of Multiple Echo Trains Using Principal Component Analysis and Independent Component Analysis |

US20090292473 * | 21. Mai 2009 | 26. Nov. 2009 | Baker Hughes Incorporated | Real-Time NMR Distribution While Drilling |

US20100010744 * | 11. Juli 2008 | 14. Jan. 2010 | Schlumberger Technology Corporation | Monte carlo method for laplace inversion of nmr data |

US20110187372 * | 1. Febr. 2011 | 4. Aug. 2011 | Baker Hughes Incorporated | Acoustic Excitation With NMR Pulse |

US20110227570 * | 22. März 2010 | 22. Sept. 2011 | Vivek Anand | Determining the larmor frequency for nmr tools |

US20130187648 * | 24. Jan. 2012 | 25. Juli 2013 | Denise E. Freed | Estimating and displaying molecular size information of a substance |

US20130265851 * | 12. März 2013 | 10. Okt. 2013 | Kees Faber | Sensor System of Buried Seismic Array |

USRE36929 * | 15. Aug. 1997 | 31. Okt. 2000 | Fujitsu Limited | Robot control system |

CN1311247C * | 13. Jan. 2004 | 18. Apr. 2007 | 施卢默格海外有限公司 | Multi-measuring NMR analysing based on maximum entropy |

CN1325940C * | 19. Nov. 1999 | 11. Juli 2007 | 施卢默格海外有限公司 | Evaluation of stratigraphic structure for well-logging using mangetic resonance |

CN100504366C | 18. Dez. 2002 | 24. Juni 2009 | 施卢默格海外有限公司 | Method for determining molecule character of hydrocarbon mixture by nuclear magnetic resonance data |

CN101328805B | 19. Juni 2008 | 26. März 2014 | 普拉德研究及开发股份有限公司 | Method and apparatus for measuring free induction decay signal and its application in composition analysis |

EP0765486A1 * | 16. Juni 1995 | 2. Apr. 1997 | Numar Corporation | Nuclear magnetic resonance determination of petrophysical properties of geologic structures |

EP0765486A4 * | 16. Juni 1995 | 19. Mai 1999 | Numar Corp | Nuclear magnetic resonance determination of petrophysical properties of geologic structures |

EP0871045A2 * | 16. März 1998 | 14. Okt. 1998 | Schlumberger Holdings Limited | Method and apparatus for measuring total nuclear magnetic resonance porosity |

EP0871045A3 * | 16. März 1998 | 30. Mai 2001 | Schlumberger Holdings Limited | Method and apparatus for measuring total nuclear magnetic resonance porosity |

WO1997034167A1 * | 14. März 1997 | 18. Sept. 1997 | Numar Corporation | Pulse sequences and interpretation techniques for nmr measurements |

WO2004017098A1 * | 19. Aug. 2003 | 26. Febr. 2004 | Baker Hughes Incorporated | Stochastic nmr for downhole use |

WO2008027491A3 * | 30. Aug. 2007 | 2. Juli 2009 | Baker Hughes Inc | Nmr echo train compression |

Klassifizierungen

US-Klassifikation | 324/303 |

Internationale Klassifikation | G01V3/32, G01R33/44 |

Unternehmensklassifikation | G01V3/32, G01N24/081 |

Europäische Klassifikation | G01N24/08A, G01V3/32 |

Juristische Ereignisse

Datum | Code | Ereignis | Beschreibung |
---|---|---|---|

2. Nov. 1992 | AS | Assignment | Owner name: SCHLUMBERGER TECHNOLOGY CORPORATION, TEXAS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNOR:FREEDMAN, ROBERT;REEL/FRAME:006489/0923 Effective date: 19921102 |

26. Apr. 1995 | AS | Assignment | Owner name: AMPEX CORPORATION, CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:AMPEX SYSTEMS CORPORATION, A DE CORPORATION;REEL/FRAME:007456/0224 Effective date: 19950426 |

7. Juli 1997 | FPAY | Fee payment | Year of fee payment: 4 |

15. Juni 2001 | FPAY | Fee payment | Year of fee payment: 8 |

3. Aug. 2005 | FPAY | Fee payment | Year of fee payment: 12 |

Drehen